A GIS-based Analysis of Hillfort Location and Morphology 9781407316079, 9781407323312

Moving away from the highly constrained, purely humanistic and empirical studies of hillfort location and morphology of

188 94 270MB

English Pages [245] Year 2018

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Front Cover
Title Page
Copyright
Table of Contents
List of Figures
1. Setting the Scene
2. Methodology and Research Questions
3. Ceredigion
4. Multi-regional GIS-based Analysisof Hillfort Location and Morphology
5. Conclusion
6. GIS Approaches and British Hillforts – LessonsLearnt and Future Prospects
Bibliography
Recommend Papers

A GIS-based Analysis of Hillfort Location and Morphology
 9781407316079, 9781407323312

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

BAR BRIT ISH SE RIE S 644

This investigation emphasises the complexity of hillforts as a class of site. It demonstrates that GIS-based analysis, when combined with fieldwork, can effectively be applied to the investigation of hillfort location and form, paving the way for future research agendas within hillforts and beyond. Jessica Murray was awarded an AHRC-funded DPhil as part of the Atlas of Hillforts in Britain and Ireland project; this book presents a revised version of the results of her thesis. She is currently an Honorary Research Associate at the University of Chester, where she is developing GIS-based methodologies to investigate site location and form.

A GIS-based Analysis of Hillfort Location and Morphology

Moving away from the highly constrained, purely humanistic and empirical studies of hillfort location and morphology of the past, this book presents a multi-regional GIS-based analysis of the form and siting of several groups of hillforts across Britain. The location and morphology of hillforts in Ceredigion, Dartmoor, Aberdeenshire, The Gower and Warminster are investigated through a combination of GIS-based analysis and field visits. An innovative approach of integrating movement with visibility is employed to investigate whether movement, visibility and topography influenced the location and morphology of these hillforts.

MURRAY

‘To the best of my knowledge this is the first time that GIS analysis of the relationship between hillfort location and morphology has been systematically applied in Britain or elsewhere in Europe.’ Dennis Harding, Abercromby Professor of Archaeology (Emeritus), University of Edinburgh

BAR 644 2018

‘By challenging previous hypotheses it usefully moves the debate forwards. … A welcome addition to the current literature on both hillfort studies and on applied GIS in archaeology.’ Dr Henry Chapman, University of Birmingham

A GIS-based Analysis of Hillfort Location and Morphology Jessica Murray BAR BRITISH SERIES 644

2018

A GIS-based Analysis of Hillfort Location and Morphology Jessica Murray BAR BRITISH SERIES 644

2018

by Published in BAR Publishing, Oxford BAR British Series A GIS-based Analysis of Hillfort Location and Morphology © Jessica Murray Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Pen Dinas as defi ned by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Dyfed Archaeological Trust 2014; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence)). UK Copyright, The Author’s moral rights under the Designs and Patents Act are hereby expressly asserted. All rights reser ved. No par t of this work may be copied, reproduced, stored, sold, distributed, scanned, saved in any for m of digital for mat or transmitted in any for m digitally, without the written per mission of the Publisher.

ISBN 9781407316079 paperback ISBN 9781407323312 e-format DOI https://doi.org/10.30861/9781407316079 A catalogue record for this book is available from the British Library

BAR titles are available from: BAR Publishing Banbury Rd, Oxford, [email protected] + ( ) + ( ) www.barpublishing.com

,

Table of Contents List of Figures ..................................................................................................................................................................... v Chapter 1 Setting the Scene .............................................................................................................................................. 1 Introduction .......................................................................................................................................................................... 1 Interpretations of the appearance of hillforts ....................................................................................................................... 1 Immigration ................................................................................................................................................................ 1 Political and social instability: defence or display? ................................................................................................... 1 A central place ........................................................................................................................................................... 2 Approaches to hillfort research ............................................................................................................................................ 3 Landscape analysis-systems theory ............................................................................................................................ 3 The role of enclosures within a hillfort system, is there a difference? ...................................................................... 4 Typologies and classification systems ........................................................................................................................ 4 Landscape analysis – hillfort location ........................................................................................................................ 7 The enhancement of earlier approaches to hillforts within the landscape ........................................................................... 8 Chapter 2 Methodology and Research Questions ..........................................................................................................11 The significance of the application of GIS-based analysis to hillfort location and morphology ....................................... 11 The significance of a multi-regional approach to hillfort location and morphology ......................................................... 11 Initial methodological stages: test area selection and the acquisition of a DTM............................................................... 11 Test area sampling .................................................................................................................................................... 11 The acquisition of Digital Terrain Models ............................................................................................................... 12 Approaches 12 Site morphology in relation to topography............................................................................................................... 12 Method of investigation............................................................................................................................................ 12 Visibility ................................................................................................................................................................... 13 Movement................................................................................................................................................................. 19 Movement AND visibility ........................................................................................................................................ 20 Conclusion. ........................................................................................................................................................................ 22 Chapter 3 Ceredigion....................................................................................................................................................... 23 Introduction ........................................................................................................................................................................ 23 Data.... ...................................................................................................................................................................... 23 Introduction to the sites ............................................................................................................................................ 23 Physical relationship of the hillfort morphology and location with the landscape topography......................................... 29 Image.................................................................................................................................................................................. 38 Correct Pathways ............................................................................................................................................................... 49 Morphologically induced correct pathways ............................................................................................................. 49 Slope-based correct pathways .................................................................................................................................. 51 Visible correct pathways .......................................................................................................................................... 54 Blind correct pathways ............................................................................................................................................. 57 The wider archaeological context ...................................................................................................................................... 73 Conclusion. ........................................................................................................................................................................ 78 Chapter 4 Multi-regional GIS-based Analysis of Hillfort Location and Morphology .............................................. 79 Introduction to the test areas .............................................................................................................................................. 79 Dartmoor .................................................................................................................................................................. 79 Aberdeenshire........................................................................................................................................................... 79 The Gower ................................................................................................................................................................ 86 Warminster ............................................................................................................................................................... 91 Summary .................................................................................................................................................................. 91 Physical relationship of the hillfort morphology and location with the landscape topography......................................... 97 Summary ................................................................................................................................................................ 114 Image................................................................................................................................................................................ 114 iii

A GIS-based Analysis of Hillfort Location and Morphology Summary ................................................................................................................................................................ 134 Correct pathways ............................................................................................................................................................. 140 Morphologically defined correct pathways ........................................................................................................... 141 Slope-based correct pathways ................................................................................................................................ 141 Visible correct pathways ....................................................................................................................................... 145 Blind correct pathways .......................................................................................................................................... 147 The wider archaeological context........................................................................................................................... 151 Summary ................................................................................................................................................................ 159 Conclusion. ...................................................................................................................................................................... 162 Chapter 5 Conclusion .................................................................................................................................................... 214 Introduction ...................................................................................................................................................................... 214 Physical relationship of hillfort morphologies and locations with the landscape topography ........................................ 214 Ostentatious and adequate responses to landscape qualities ........................................................................................... 215 Topographic and morphological prominence .................................................................................................................. 216 Visual prominence and the definition of hillforts ............................................................................................................ 217 The falsification of images through disproportionate vallation sequences ...................................................................... 217 Visually open and closed hillforts .................................................................................................................................... 218 Correct pathways ............................................................................................................................................................. 219 Beyond movement, visibility and topography ................................................................................................................. 220 Physical associations with the past......................................................................................................................... 220 Wider chronological picture ................................................................................................................................... 220 Concluding remarks ......................................................................................................................................................... 221 Chapter 6 GIS Approaches and British Hillforts – Lessons Learnt and Future Prospects .................................... 222 Revisiting the aims and objectives ................................................................................................................................... 222 A critical analysis of the methodology and data .............................................................................................................. 222 Does the application of GIS-based analysis enhance Driver’s non-GIS-based analysis? ............................................... 223 The physical relationship between hillfort morphology and topography ........................................................................ 223 Image................................................................................................................................................................................ 223 Correct pathways ............................................................................................................................................................. 224 The step forward and future research prospects ............................................................................................................. 225 Bibliography ................................................................................................................................................................... 226

iv

List of Figures Figure 1 Distribution density map of Hillforts in Britain .................................................................................................... 5 Figure 2 Result of investigations into the effect that inner bank height has upon the visibility of Hardings Down North enclosure from The Bulwark ........................................................................................................................ 14 Figure 3 Result of investigations into the effect that inner bank height has upon the visibility of Scratchbury from Cley Hill ................................................................................................................................................................... 15 Figure 4 Schematic explanation of VG compared to VD .................................................................................................. 17 Figure 5 Comparison of the effectiveness of 25, 49 and 100 grid point viewsheds, depicting the visual magnitude of Moel Arthur ................................................................................................................................................................... 18 Figure 6 Calculating the blind pathway friction surface .................................................................................................... 21 Figure 7 Calculating the visible pathway friction surface ................................................................................................. 21 Figure 8 Explanation of the friction surface for calculating a routeway where architectural components are most visible within the landscape ...................................................................................................................................... 22 Figure 9 Distribution of hillforts within the Ceredigion test area ..................................................................................... 24 Figure 10 1m resolution LiDAR hillshade models of Pen y Bannau and its immediate environs .................................... 25 Figure 11 1m resolution LiDAR hillshade models of Castell Tregaron and its immediate environs ................................ 26 Figure 12 1m resolution LiDAR hillshade models of Trecoll and its immediate environs ............................................... 27 Figure 13 1m resolution LiDAR hillshade models of Gaer Fawr and its immediate environs ......................................... 28 Figure 14 1m resolution LiDAR hillshade models of Castell Grogwynion and its immediate environs .......................... 30 Figure 15 1m resolution LiDAR hillshade models of Tan y Ffordd and its immediate environs ...................................... 31 Figure 16 50cm resolution LiDAR hillshade models of Pen Dinas and its immediate environs ...................................... 32 Figure 17 1m resolution LiDAR hillshade models of Pen Dinas Elerch and its immediate environs ............................... 33 Figure 18 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Castell Tregaron as defined by the hillfort buffers ............................................................................................................. 34 Figure 19 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Pen y Bannau as defined by the hillfort buffers ................................................................................................................. 35 Figure 20 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Trecoll as defined by the hillfort buffers ............................................................................................................................ 36 Figure 21 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Pen Dinas as defined by the hillfort buffers ....................................................................................................................... 37 Figure 22 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Gaer Fawr as defined by the hillfort buffers ...................................................................................................................... 38 Figure 23 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Tan y Ffordd as defined by the hillfort buffers................................................................................................................... 39 Figure 24 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Pen Dinas Elerch as defined by the hillfort buffers............................................................................................................ 40 Figure 25 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Castell Grogwynion as defined by the hillfort buffers ....................................................................................................... 41 Figure 26 Results of viewshed analysis from the radii towards Castell Grogwynion depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ............................................................................................... 42

v

A GIS-based Analysis of Hillfort Location and Morphology Figure 27 Results of viewshed analysis indicating the visibility of Castell Grogwynion from Pen y Bannau ................. 43 Figure 28 Results of viewshed analysis from the radii towards Castell Tregaron depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ..................................................................................................... 45 Figure 29 Results of viewshed analysis from the radii towards Tan y Ffordd depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ........................................................................................................... 46 Figure 30 Results of viewshed analysis from the radii towards Trecoll depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ..................................................................................................................... 47 Figure 31 Results of viewshed analysis from the radii towards Pen Dinas Elerch depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ..................................................................................................... 48 Figure 32 Results of viewshed analysis from the radii towards Pen y Bannau depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ........................................................................................................... 50 Figure 33 Results of viewshed analysis indicating the visibility of Pen y Bannau from Castell Grogwynion ................. 51 Figure 34 Results of viewshed analysis from the radii towards Gaer Fawr depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ........................................................................................................... 52 Figure 35 Visibility of Gaer Fawr from Pen Dinas. Map A depicts the viewshed results when using the Ordnance Survey PROFILE DTM, Map B depicts the results when using the 1m resolution LiDAR............................. 53 Figure 36 Visibility of Pen Dinas from Gaer Fawr. Map A depicts the viewshed results when using the Ordnance Survey PROFILE DTM, Map B depicts the results when using the 1m resolution LiDAR............................. 54 Figure 37 Results of viewshed analysis from the radii towards Pen Dinas depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ........................................................................................................... 55 Figure 38 Results of slope-based cost surface analysis at Castell Grogwynion ................................................................ 56 Figure 39 Results of slope-based cost surface analysis at Tan y Ffordd............................................................................ 57 Figure 40 Results of slope-based cost surface analysis at Pen Dinas Elerch..................................................................... 58 Figure 41 Results of slope-based cost surface analysis at Castell Tregaron ...................................................................... 59 Figure 42 Results of slope-based cost surface analysis at Gaer Fawr ............................................................................... 60 Figure 43 Results of slope-based cost surface analysis at Pen y Bannau .......................................................................... 61 Figure 44 Results of slope-based cost surface analysis at Trecoll ..................................................................................... 62 Figure 45 Results of slope-based cost surface analysis at Pen Dinas ................................................................................ 63 Figure 46 Distribution of visible pathways at Trecoll ....................................................................................................... 64 Figure 47 Distribution of visible pathways at Castell Grogwynion .................................................................................. 65 Figure 48 Distribution of visible pathways at Gaer Fawr .................................................................................................. 66 Figure 49 Distribution of visible pathways at Pen Dinas Elerch ....................................................................................... 67 Figure 50 Distribution of visible pathways at Tan y Ffordd .............................................................................................. 68 Figure 51 Distribution of visible pathways at Pen y Bannau............................................................................................. 69 Figure 52 Distribution of visible pathways at Pen Dinas .................................................................................................. 70 Figure 53 Distribution of blind pathways at Gaer Fawr .................................................................................................... 71 Figure 54 Distribution of blind pathways at Pen y Bannau ............................................................................................... 72 Figure 55 Distribution of blind pathways at Castell Tregaron........................................................................................... 73 Figure 56 Distribution of blind pathways at Trecoll .......................................................................................................... 74 Figure 57 Distribution of blind pathways at Pen Dinas Elerch ......................................................................................... 75 Figure 58 Distribution of blind pathways at Tan y Ffordd ................................................................................................ 76 Figure 59 Results of cost surface analysis indicating the correlation between the least cost pathways and known areas of activity at Pen Dinas Elerch ..................................................................................................................... 77

vi

List of Figures Figure 60 Distribution of hillforts within the Dartmoor test area ...................................................................................... 80 Figure 61 Hillshade model of 1m resolution LiDAR DTM for Cranbrook Castle ............................................................ 81 Figure 62 Hillshade model of 1m resolution LiDAR DTM for Wooston Castle ............................................................... 82 Figure 63 Hillshade model of 1m resolution LiDAR DTM for Prestonbury Castle ......................................................... 83 Figure 64 Distribution of hillforts within the Aberdeenshire test area .............................................................................. 84 Figure 65 Plan of Dunnideer .............................................................................................................................................. 85 Figure 66 Plan of Tap o’ Noth ............................................................................................................................................ 85 Figure 67 Plan of Wheedlemont ........................................................................................................................................ 86 Figure 68 Plan of Cairnmore.............................................................................................................................................. 86 Figure 69 Distribution of hillforts within the Gower test area........................................................................................... 87 Figure 70 Hillshade model of 2m resolution LiDAR DTM for The Bulwark ................................................................... 88 Figure 71 Hillshade model of 2m resolution LiDAR DTM for Hardings Down West Enclosure..................................... 89 Figure 72 Hillshade model of 2m resolution LiDAR DTM for Hardings Down North Enclosure ................................... 90 Figure 73 Hillshade model of 2m resolution LiDAR DTM for Hardings Down East Enclosure ..................................... 92 Figure 74 Distribution of hillforts within the Warminster test area ................................................................................... 93 Figure 75 Hillshade model of 1m resolution LiDAR DTM for Cley Hill ......................................................................... 94 Figure 76 Hillshade model of 1m resolution LiDAR DTM for Battlesbury ..................................................................... 95 Figure 77 Hillshade model of 1m resolution LiDAR DTM for Scratchbury .................................................................... 96 Figure 78 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Cranbrook Castle as defined by the hillfort buffers ........................................................................................................... 98 Figure 79 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Wheedlemont as defined by the hillfort buffers ................................................................................................................. 99 Figure 80 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Cley Hill as defined by the hillfort buffers....................................................................................................................... 100 Figure 81 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Scratchbury as defined by the hillfort buffers .................................................................................................................. 101 Figure 82 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Battlesbury as defined by the hillfort buffers ................................................................................................................... 103 Figure 83 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Cairnmore as defined by the hillfort buffers .................................................................................................................... 104 Figure 84 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Dunnideer as defined by the hillfort buffers .................................................................................................................... 105 Figure 85 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Tap o’Noth as defined by the hillfort buffers ................................................................................................................... 106 Figure 86 View from of Prestonbury from Broadmoor Common ................................................................................... 107 Figure 87 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Prestonbury Castle as defined by the hillfort buffers ....................................................................................................... 108 Figure 88 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Wooston Castle as defined by the hillfort buffers ............................................................................................................ 109 Figure 89 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from The Bulwark as defined by the hillfort buffers ................................................................................................................ 110 Figure 90 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Hardings Down West Enclosure as defined by the hillfort buffers ...................................................................................111

vii

A GIS-based Analysis of Hillfort Location and Morphology Figure 91 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Hardings Down North Enclosure as defined by the hillfort buffers................................................................................. 112 Figure 92 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Hardings Down East Enclosure as defined by the hillfort buffers ................................................................................... 113 Figure 93 Results of cost surface analysis highlighting the routes where Prestonbury’s enclosing works are most visible ...................................................................................................................................................................... 115 Figure 94 Results of viewshed analysis indicating the visibility of Prestonbury Castle from Wooston Castle .............. 116 Figure 95 Results of viewshed analysis indicating the visibility of Prestonbury Castle from Cranbrook Castle ........... 117 Figure 96 Results of viewshed analysis from the radii towards Prestonbury depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ........................................................................................................ 118 Figure 97 Visibility of Prestonbury Castle from the surrounding landscape ................................................................... 119 Figure 98 Results of cost surface analysis highlighting the routes where Battlesbury’s enclosing works are most visible ..................................................................................................................................................................... 120 Figure 99 Results of viewshed analysis indicating the visibility of Battlesbury from Cley Hill..................................... 121 Figure 100 View of Battlesbury from Cley Hill .............................................................................................................. 121 Figure 101 Results of viewshed analysis indicating the visibility of Battlesbury from Scratchbury .............................. 122 Figure 102 View of Battlesbury from Scratchbury .......................................................................................................... 122 Figure 103 Results of viewshed analysis from the radii towards Battlesbury depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ......................................................................................................... 123 Figure 104 Visibility of Battlesbury from the surrounding landscape ............................................................................. 124 Figure 105 Results of cost surface analysis highlighting the routes where The Bulwark’s enclosing works are most visible ...................................................................................................................................................................... 125 Figure 106 Results of viewshed analysis indicating the visibility of The Bulwark from Hardings Down North Enclosure ............................................................................................................................................................... 126 Figure 107 Results of viewshed analysis indicating the visibility of The Bulwark from Hardings Down West Enclosure................................................................................................................................................................. 127 Figure 108 Results of viewshed analysis indicating the visibility of The Bulwark from Hardings Down East Enclosure.................................................................................................................................................................. 128 Figure 109 View of The Bulwark from Hardings Down ................................................................................................. 129 Figure 110 Results of viewshed analysis from the radii towards The Bulwark depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ................................................................................................... 130 Figure 111 View of The Bulwark from the surrounding landscape ................................................................................. 131 Figure 112 Results of cost surface analysis highlighting the routes where Cley Hill’s enclosing works are most visible ...................................................................................................................................................................... 132 Figure 113 Results of viewshed analysis indicating the visibility of Cley Hill from Battlesbury ................................... 133 Figure 114 Results of viewshed analysis indicating the visibility of Cley Hill from Scratchbury .................................. 134 Figure 115 Results of viewshed analysis from the radii towards Cley Hill depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ......................................................................................................... 135 Figure 116 View of Cley Hill from the surrounding landscape ....................................................................................... 136 Figure 117 Results of viewshed analysis from the radii towards Cranbrook Castle depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ............................................................................................. 137 Figure 118 View of Cranbrook Castle from the surrounding landscape .......................................................................... 138 Figure 119 Results of viewshed analysis indicating the visibility of Cranbrook Castle from Prestonbury Castle ......... 139 Figure 120 Results of cost surface analysis highlighting the routes where Cranbrook Castle’s enclosing works are most visible ................................................................................................................................................................ 140 viii

List of Figures Figure 121 Results of viewshed analysis indicating the visibility of Cranbrook Castle from Wooston Castle............... 141 Figure 122 Results of viewshed analysis from the radii towards Wheedlemont depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ................................................................................................... 142 Figure 123 View of Wheedlemont from Cairnmore ........................................................................................................ 143 Figure 124 Results of viewshed analysis indicating the visibility of Wheedlemont from Cairnmore ............................ 144 Figure 125 View of Wheedlemont from Tap o’Noth ....................................................................................................... 145 Figure 126 Results of viewshed analysis indicating the visibility of Wheedlemont from Tap o’Noth ........................... 146 Figure 127 Results of viewshed analysis indicating the visibility of Wheedlemont from Dunnideer............................. 147 Figure 128 View of Wheedlemont from the surrounding landscape ............................................................................... 148 Figure 129 Viewshed results indicating the visibility of Cairnmore from the surrounding radii .................................... 149 Figure 130 Results of viewshed analysis indicating the visibility of Cairnmore from Wheedlemont ............................ 150 Figure 131 View of Cairnmore from Wheedlemont ........................................................................................................ 151 Figure 132 Results of viewshed analysis indicating the visibility of Cairnmore from Tap o’Noth ................................ 152 Figure 133 Results of viewshed analysis indicating the visibility of Cairnmore from Dunnideer .................................. 153 Figure 134 Results of viewshed analysis indicating the visibility of Dunnideer from the surrounding radii ................. 154 Figure 135 View of Dunnideer from the surrounding landscape ..................................................................................... 155 Figure 136 Viewshed results indicating the visibility of Dunnideer from Tap o’Noth .................................................... 156 Figure 137 Results of viewshed analysis indicating the visibility of Dunnideer from Wheedlemont............................. 157 Figure 138 Results of viewshed analysis indicating the visibility of Dunnideer from Cairnmore .................................. 158 Figure 139 Results of viewshed analysis indicating the visibility of Tap o’Noth from the surrounding radii when using the Ordnance Survey PROFILE DTM ......................................................................................................... 159 Figure 140 View of Tap o’Noth from the surrounding landscape ................................................................................... 160 Figure 141 Results of viewshed analysis indicating the visibility of Tap o’Noth from Dunnideer when using the Ordnance Survey PROFILE DTM ............................................................................................................................. 161 Figure 142 View of Tap o’Noth from Dunnideer............................................................................................................. 162 Figure 143 Results of viewshed analysis indicating the visibility of Tap o’Noth from Cairnmore when using the Ordnance Survey PROFILE DTM ............................................................................................................................. 163 Figure 144 Results of viewshed analysis indicating the visibility of Tap o’Noth from Wheedlemont when using the Ordnance Survey PROFILE DTM ................................................................................................................... 164 Figure 145 View of Tap o’Noth from Wheedlemont ....................................................................................................... 165 Figure 146 View of Tap o’Noth from Cairnmore ............................................................................................................ 165 Figure 147 Results of viewshed analysis from the radii towards Hardings Down North Enclosure depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ................................................................... 166 Figure 148 View of Hardings Down North Enclosure from the surrounding landscape ................................................. 167 Figure 149 Results of viewshed analysis indicating the visibility of Hardings Down North Enclosure from Hardings Down West Enclosure ...................................................................................................................................... 168 Figure 150 Results of viewshed analysis indicating the visibility of Hardings Down North Enclosure from Hardings Down East Enclosure ....................................................................................................................................... 169 Figure 151 Results of viewshed analysis indicating the visibility of Hardings Down North Enclosure from The Bulwark..................................................................................................................................................................... 170 Figure 152 View of Hardings Down North Enclosure from The Bulwark ...................................................................... 170 Figure 153 Results of viewshed analysis from the radii towards Hardings Down East Enclosure depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ......................................................................... 171

ix

A GIS-based Analysis of Hillfort Location and Morphology Figure 154 Results of viewshed analysis indicating the visibility of Hardings Down East Enclosure from Hardings Down North Enclosure ..................................................................................................................................... 172 Figure 155 Results of viewshed analysis indicating the visibility of Hardings Down East Enclosure from Hardings Down West Enclosure ...................................................................................................................................... 172 Figure 156 Results of viewshed analysis indicating the visibility of Hardings Down East Enclosure from The Bulwark..................................................................................................................................................................... 173 Figure 157 View of Hardings Down East Enclosure from The Bulwark ........................................................................ 173 Figure 158 Results of cost surface analysis highlighting the routes where Hardings Down East Enclosure’s enclosing works are most visible ..................................................................................................................................... 174 Figure 159 Results of viewshed analysis from the radii towards Hardings Down West Enclosure depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ......................................................................... 175 Figure 160 View of Hardings Down West Enclosure from the surrounding landscape .................................................. 176 Figure 161 Results of viewshed analysis indicating the visibility of Hardings Down West Enclosure from Hardings Down East Enclosure ....................................................................................................................................... 177 Figure 162 View of Hardings Down West Enclosure from Hardings Down East Enclosure .......................................... 177 Figure 163 Results of viewshed analysis indicating the visibility of Hardings Down West Enclosure from Hardings Down North Enclosure ..................................................................................................................................... 178 Figure 164 Results of viewshed analysis indicating the visibility of Hardings Down West Enclosure from The Bulwark..................................................................................................................................................................... 178 Figure 165 Results of cost surface analysis highlighting the routes where Hardings Down West Enclosure’s enclosing works are most visible ..................................................................................................................................... 179 Figure 166 Results of viewshed analysis from the radii towards Wooston Castle depicting the visibility of the site when using the Ordnance Survey PROFILE DTM ................................................................................................... 180 Figure 167 Results of viewshed analysis indicating the visibility of Wooston Castle from Cranbrook Castle............... 181 Figure 168 Results of viewshed analysis indicating the visibility of Wooston Castle from Prestonbury Castle ............ 181 Figure 169 Results of cost surface analysis highlighting the routes where Wooston Castle’s enclosing works are most visible ................................................................................................................................................................ 182 Figure 170 Results of slope-based cost surface analysis indicating the accessibility of Dunnideer ............................... 183 Figure 171 Results of slope-based cost surface analysis indicating the accessibility of Cairnmore ............................... 184 Figure 172 Results of slope-based cost surface analysis indicating the accessibility of Wooston Castle ....................... 185 Figure 173 Results of slope-based cost surface analysis indicating the accessibility of Prestonbury Castle .................. 185 Figure 174 Results of slope-based cost surface analysis indicating the accessibility of Tap o’Noth .............................. 186 Figure 175 Results of slope-based cost surface analysis indicating the accessibility of Wheedlemont .......................... 187 Figure 176 Results of slope-based cost surface analysis indicating the accessibility of Cranbrook Castle .................... 188 Figure 177 Results of slope-based cost surface analysis indicating the accessibility of Hardings Down West Enclosure................................................................................................................................................................. 188 Figure 178 Results of slope-based cost surface analysis indicating the accessibility of Hardings Down East Enclosure.................................................................................................................................................................. 189 Figure 179 Results of slope-based cost surface analysis indicating the accessibility of Cley Hill ................................. 190 Figure 180 Results of slope-based cost surface analysis indicating the accessibility of Battlesbury .............................. 191 Figure 181 Sum of visible least cost pathways at Cranbrook Castle ............................................................................... 192 Figure 182 Sum of visible least cost pathways at Wooston Castle .................................................................................. 193 Figure 183 Sum of visible least cost pathways at Prestonbury Castle ............................................................................. 194 Figure 184 Sum of visible least cost pathways at Dunnideer .......................................................................................... 195

x

List of Figures Figure 185 Sum of visible least cost pathways at Scratchbury ........................................................................................ 196 Figure 186 Sum of visible least cost pathways at Battlesbury......................................................................................... 197 Figure 187 Sum of visible least cost pathways at the Bulwark ....................................................................................... 198 Figure 188 Sum of blind least cost pathways at Hardings Down North Enclosure ......................................................... 199 Figure 189 Sum of blind least cost pathways at Battlesbury ........................................................................................... 200 Figure 190 Sum of blind least cost pathways at Hardings Down West Enclosure .......................................................... 201 Figure 191 Sum of blind least cost pathways at Cairnmore ............................................................................................ 202 Figure 192 Sum of blind least cost pathways at Cley Hill ............................................................................................... 203 Figure 193 Sum of blind least cost pathways at Wheedlemont ....................................................................................... 204 Figure 194 Landscape wide distribution of visible least cost pathways to and from Wheedlemont ............................... 205 Figure 195 Landscape wide distribution of visible least cost pathways to and from Dunnideer and their correlation with known archaeology ................................................................................................................................ 206 Figure 196 Landscape wide distribution of visible least cost pathways to and from Tap o’Noth ................................... 207 Figure 197 Landscape wide distribution of slope-based least cost pathways to and from Dunnideer ............................ 208 Figure 198 Landscape wide distribution of slope-based least cost pathways to and from Wheedlemont....................... 209 Figure 199 Landscape wide distribution of slope-based least cost pathways to and from Tap o’Noth ........................... 210 Figure 200 Landscape wide distribution of slope-based least cost pathways to and from Cairnmore ............................ 211 Figure 201 Landscape wide distribution of visible least cost pathways to and from Wooston Castle and its correlation with known archaeology ................................................................................................................................ 212 Figure 202 Landscape wide distribution of visible least cost pathways to and from Cranbrook Castle and its correlation with known archaeology ................................................................................................................................ 213

xi

1 Setting the Scene Introduction

the archaeological evidence as opposed to interpreting the evidence alone.

This study tests the applicability and effectiveness of applying GIS-based analytical techniques within investigations into the morphology and topographical location of a sample of hillforts in Britain. This chapter sets out the study’s context by discussing past approaches to hillforts. It begins by detailing how ideas surrounding their appearance has changed. This is followed by an exploration of the key approaches to the examination of hillforts for example through Central Place Theory, typological classifications and investigations focusing on assessing their physical position within the landscape. This background is concluded by introducing and examining the key influences, approaches and aims of this research.

Four decades later, Hogg analysed the distribution of hillforts along the Welsh coast and argued that their size, form and shape failed to display a clear sense of identity (1972). However, he suggested that their distribution may reflect the spread of people arriving at various points along the coast of Wales (Ibid., 12). Similarly, Chitty assessed the location and date of objects within the Welsh Marches to investigate how and when the people that constructed the hillforts came into the area (1937). Both Hogg and Chitty analysed the archaeological evidence alone. However, by taking into account the distribution and form of various types of sites and material culture they were able to begin to question its wider social implications. This approach was motivated by the foci of contemporary studies which assessed population movement, and is explored below.

This work was AHRC funded and is part of the Atlas of Hillforts in Britain and Ireland Project; one aspect of this focuses on assessing concepts of regionality within hillforts across England, Wales, Scotland, Northern Ireland and Ireland (2015). As will be explained in greater detail in Chapter 2, this work does not attempt to identify patterns of location which can be categorised as ‘national’, i.e. typical of Scotland or Wales, as the samples used are small and not representative of the variability between hillforts within any one country. Rather, it utilises the testing of a methodology within a variety of test areas (again, detailed in Chapter 2) as an opportunity to explore the characteristics of topographical setting and the morphological characteristics of hillforts. This enables an investigation into whether there are any similarities and differences which may be meaningful within the wider context of hillforts and their surrounding landscape.

Political and social instability: defence or display? Interpretations based on population movement began to decline and new reasons for the appearance and development of hillforts were suggested. For example, Simpson maintained the belief that when hillforts were constructed there was a mass movement of people (1964). She saw this as an invasion that caused the natives to construct hillforts as a defensive mechanism (1964). However, climate deterioration between the Bronze Age and Iron Age exacerbated land deterioration (Champion 1999, 103), consequently good land was sought after and there was a need to define space (Moore 2007, 274). Some interpretations revolved around stronger feelings of insecurity, these suggested that hillforts actually originated from a need for defence. Their origins were interpreted as a response to a defensive need and their defensive function defined the site (Sutton 1966). Similarly, Dyer saw hillforts as a military construction which offered the ‘best protection possible to its inhabitants’ (1992, 5).

Interpretations of the appearance of hillforts Immigration Early hillfort studies focused upon interpreting individual site function and origins. The political climate at the beginning of the twentieth century greatly influenced the directions taken within hillfort studies and this can be seen within Hawkes’ work in particular (1931). Hawkes saw hillforts to be the result of conflict due to threats from Celtic immigration (1931). He believed that hillforts protected ‘country folk . . . [and] their stock’ (1931, 76). His work was also influenced by classical literature as he talked a lot about the movement of tribes and showed that such detailed information could only be retrieved from such writings (Ibid.). It is now generally accepted that these writings focused upon military and political subjects (James & Rigby 1997, 4-5), they formed militaristic propaganda. Hawkes used these writings together with

For some, morphology defined the defensive function of hillforts. These interpretations imply that hillforts were a design that was based upon a functional need, a design that was enforced upon an area as a result of external stimulation. For example, Hogg’s definition of hillforts was based upon sites which had ‘substantial defences, usually on high ground and probably built between about 1000 B.C. and A.D. 700’ (1979, 1). The topographic and defensive definition of hillforts was also supported by Bray and Tramp who defined a hillfort as a ‘fortified hilltop’ (1970, 104). Bowden’s interpretation argues that the name, form and location of these sites suggest that they are by 1

A GIS-based Analysis of Hillfort Location and Morphology to these structures were initially designed to impress but by the first century BC they formed a defensive function (Ibid., 66).

definition a defended place (2011, 2). Fox also argued that the form of a hillfort’s enclosing earthworks gave it a defensive function, in particular she argued that the sites with widely spaced ramparts were defensive (1961).

However, Westropp had much earlier argued that the enclosing works to a hillfort were an accessory and not a necessity (1896/1901, 638). The non-defensive interpretation of banks and ditches has increasingly been adopted by researchers. For example, Harding argues that multivallation is not necessarily evidence for hillfort development, it may have been used as a status symbol to impress (2012, 13). Hillforts were not necessarily defensive, similarly the expenditure upon enclosing works was not always directly proportional to the defensive need to enclose (Ibid.,13). Hamilton and Manley also argued that the enhancement of hillforts and the delineation of space outside of them was an indicator of competition between hillforts rather than defensive (2001).

On the contrary, other scholars saw the location of hillforts to aid and sometimes define their function. For example, Avery believed that hillforts were distinguished by their height advantage over those who approached them; consequently, he argued that the term ‘hillfort’ should only be used for sites with this advantage (1976, 4). This view was long standing, for example, Anderson also took location into account but he saw hillforts as an adaptation of an already elevated, defensive location (1883, 271). Similarly, Harding, in the most recent account of hillforts, noted that they may have been ‘tactically’ positioned to take advantage of naturally defended and visually commanding positions which are also located close to resources (2012, 15).

A non-utilitarian interpretation of hillfort enclosure was also put forward by Hingley who argued that their banks and ditches potentially distinguished insiders from outsiders so that they were about inclusion and exclusion (1990). This interpretation was also reflected within Richard Bradley’s work which saw hillforts as public monuments (2005). Harding also acknowledged that hillforts may have been positioned to be visually prominent (2012, 15), which was also argued by Sharples (2010) and Cunliffe (2006). The creation of such visually prominent monuments could have been in an attempt to legitimise the status and/or social position (Parker Pearson 1984, 71) of those that were within the site. Kelly argues that there should be a dichotomy between ‘public and restricted performance sites’ (2015, 27), however in the case of hillforts the writer wishes to investigate whether this is visibly the case. The process by which the visibility of the enclosing works of hillforts compared to their interiors provides an insight into the public and/or private nature of hillforts is explored later in this book.

The defensive interpretation of hillfort function was not adopted by everyone. Very early in Irish hillfort studies, Westropp argued that the enclosing banks and ditches of ‘Celtic forts’ were a ‘passive defence for these houses, and [were] only raised against a sudden attack, not against undermining, battering or other siege work’ (1896/1901, 638). Similarly, Crawford and Keiller argued that the enclosing earthworks at sites such as Hod Hill ‘were designed to repel invaders, not to stand a siege’ (1928, 8). This interpretation was maintained by Bowden and McOmish who highlighted that during the Iron Age, there were underlying tensions that meant that there may have been militaristic motivations, but sites were developed to repel aggression (1987). They also argued that hillforts formed centres for a ‘detached elite’ and enabled them to legitimise their place within the social system as opposed to offering a safe place (Ibid.). However, Lock argued that interpretations which relate to the appearance and maintenance of hillforts need to move away from giving the impression of a period of endemic warfare (2011). He based his interpretation on the fact that there is limited evidence for warfare across Britain, although he admits that hillforts may still have resulted from feelings of insecurity (Ibid.). Instead, based upon the rise in defining communities through enclosure, Lock argues that cosmological threats were seen to risk the cohesion of the social group (Ibid.). Activities and places were created to enhance and protect social cohesion, an example of such an activity could be the maintenance of agricultural land and the construction and maintenance of hillforts (Ibid.).

A central place Although there was a great deal of variation within interpretations as to why and how hillforts began to be constructed, these sites were predominantly interpreted as ‘central places’ i.e. they served a wider, dispersed community. However, interpretations surrounding the extent and role of these central places have varied greatly. As a place for community activities, it was believed that their role as a central place may have varied over time. For example, Collis argued that hillforts are evidence of the centralisation of defence (1994a, 34; 1994b, 131) and redistribution (Ibid. 1994b, 131). Similarly, Kӧhler saw them as acting as defence for a community with their location also being advantageous for access to resources (1995, 165).

Similarly, Fox recognised that hillforts were used in times of peace (1961). During these times, the morphology of the enclosing earthworks had a social and ceremonial function as they provided ‘an approach of dignity to the principal enclosure’ (Fox 1961 45-46). Cunliffe saw the definition and form of hillforts to have changed over time in relation to defensive need (1971). According to Cunliffe, the gates

Interpretations of hillforts being central places went beyond seeing them as places that satisfied a practical and functional need towards seeing them as a place of social interaction. For 2

Setting the Scene example, in the 19th century, Westropp saw Irish hillforts to have acted as places of assembly such as a church or a place of worship; however, he also postulated that they could have been cattle enclosures (1896/1901, 637). Much later, Harding amongst others, also saw hillforts as places of assembly for ritual or social purposes (2012, 282). According to Barclay and others ‘the hillfort created an obvious and visible site and was also a focus of communal action which bound people together’ (2003, 250). As Sharples argued, the construction of enclosures such as hillforts defined the relationships of the people that were involved within this process (2010). The construction of hillforts was a change, a change of site form; which was inevitably caused by wider social changes. Community is spatially defined (Lock and Gosden 2005, 133), in this case they were potentially defined, on one level, by the walls of the hillfort. These communities arose from ‘historically embedded relationships and are nurtured through encountering and reacting to new situations and people’ (Ibid.).

Williams equated enclosure size with importance and argued that size was a direct reflection of the amount of surplus which came from the surrounding landscape (1988). This surplus, and subsequently the size of the site, was affected by the quality and type of land that was within the settlement zone (Ibid.). The correlation of site size with site importance was argued earlier by Clarke who stated that within a ‘fully developed, Celtic settlement hierarchy’ the largest sites (oppida) provided the highest order goods and services (1972, 864). Equating a site hierarchy with a social hierarchy became influential in Cunliffe’s later interpretation of hillforts and enclosures in Hampshire. Here he argued that the size, siting and complexity of hillforts implied that they were built under coercion. This set them aside from other settlements as the construction of such a site would have needed a large group of people (1972; 1978; 1991). Another big influence upon hillfort studies in the 1970s was systems theory. This suggested that to understand an object or a site, an understanding of the cultural system within which it sits needs to be established (Clarke 1978). This is a ‘unit system in which all the cultural information is a stabilized but constantly changing network of intercommunicating attributes forming a complex wholea dynamic system’ (Ibid., 42). The need to understand systems through a series of single sites was suggested by Hodges’ who recognised that to be able to identify the trade networks of hillforts within Ireland there needed to be more excavations (1975).

Approaches to hillfort research Landscape analysis-systems theory Central Place Theory was introduced by Christaller in 1933 (Ullman 2005) and was popular in archaeology during the 1970s. It worked upon the basis of a ‘functional interdependence between a town and the surrounding rural area’ (King 1984, 29). This form of economic modelling was used by Clarke to assess the distribution of sites surrounding Cadbury hillfort in Somerset (1972). Through modelling the settlement pattern, Clarke postulated a landscape scale site hierarchy, with Cadbury being the central place that was surrounded by a series of dependent settlements and farmsteads (Ibid.). Also on a hierarchical basis, Cunliffe saw hillforts as home to ruling elite (1972; 1978). On a socio-economic basis, Cunliffe also postulated that the elite also lived in the ‘large farmsteads of Little Woodbury type’ and those that they ruled over brought surplus to the hillforts which acted as redistribution centres and areas for group gatherings (Ibid.,333).

Systems Theory influenced research programmes. At Danebury for example, to question the ‘system’ and the site’s function Cunliffe employed a large sampling strategy (Cunliffe 1984; Cunliffe and Poole 1991). Within the first series of excavations (1969-1978), focus was put upon sampling as much of the site as possible to establish a site function and construction sequence (Cunliffe 1984). The excavation was accompanied by an aerial photographic interpretation of the landscape around Danebury; this depicted the settlement pattern (Palmer 1984). Subsequent excavation seasons (1979-1988) continued the extensive process of sampling to establish a site chronology at Danebury (Cunliffe and Poole, 1991). This work provided the basis to understanding Danebury’s function within its wider landscape context. Danebury is one of the country’s most extensively excavated hillforts and provides a dataset where questions of functionality could be feasibly answered. Cunliffe acknowledged this potential and continued to try to establish the function of Danebury within his latter Danebury Environs publication where he also aimed ‘to forward our knowledge of the organization of the landscape in the first millennium BC’ (2000, 14)

Still in the 1970s, Hogg recognised the difficulty in defining territories with such a limited number of sites that were positioned in clearly defined land units (1971). He therefore tested quantitative methods of calculating territories (Ibid.). The most popular method was Thiessen Polygons. This modelled site territories, and weighted distances between the hillforts and their territorial boundaries (Ibid.). However, using models to define territory is not realistic, as if territories did exist during the Iron Age, they would not have been calculated through modelling. The land units associated with hillforts, for example, could have been defined by the land’s ability to provide the community with the required resources, Cunliffe defined this as a ‘zone of exploitation’ (1991, 24).

Although there has been an extensive investment into the investigation of Danebury and its landscape, Cunliffe acknowledges that any ‘system’ in one area was not necessarily the same as in another (2001). Regardless of whether scholars acknowledged the potential for crossregional variation in social systems, designing research

Regardless of their unrealistic expectations, the application of Thiessen Polygons to hillfort research encouraged the equation of enclosure size with territories (Hogg 1971). 3

A GIS-based Analysis of Hillfort Location and Morphology projects to investigate system mechanics as opposed to questioning their existence immediately implies that systems were indeed in place. This also imposes a system and rigidity upon the archaeological record that may not have existed within the past.

cultural affinities which had been gleamed from material culture derived from excavations (Rivet 1961, 30-31). Similarly, Piggott developed a classification of Scottish hillforts which was based on the cultural affinities of those who constructed them (1966).

The role of enclosures within a hillfort system, is there a difference?

Chitty’s map of the Iron Age B south-western culture plotted ‘camps’ based upon their entrance type (1938). This typified the increasing focus upon hillfort typologies, for example, although the map of Southern Iron Age Britain was primarily based upon Iron Age cultures it also plotted the size and vallation of the hillforts (Rivet 1961). The vallation categories were divided into multivallate and univallate whilst site size was categorised as: under 3 acres, 3-15 acres and 15+ acres (Ibid. 1961).

Central Place Theory influenced the socio-economic interpretation of hillforts as a comparison to other enclosures. Cunliffe saw hillforts as a level of social organisation that was above a farmstead or a hamlet that were potentially separate phenomena (1991, 312). Cunliffe also argued that the presence of these substantially ‘different’ sites created social difference (Ibid.), similarly Champion believed that hillforts were a means to social differentiation due to their economic role (1994, 133). According to Stanford, hillforts were a different class to enclosures because of the density of buildings and the types of houses that were in them (1971, 48).

Forde Johnston also defined hillforts on a morphological and locational basis (1976), he identified seven location types which were; • Hill-top situations • Promontory and semi-contour situations (‘sites in which there is an easy, or relatively easy approach on one side, the other three sides being defended either completely or in part by nature’), • promontory and semi-contour situations (‘the approach to the site is on a front which accounts for roughly half the circuit’) • ridge-top situations (‘embraces sites in which there is an easy or relatively easy approach on two sides’) • cliff-edge and plateau-edge situations (‘there is a level, or at least an easy, approach on three sides with natural defences only on the fourth side’) • Hill-slope situations • plateau and low-lying situations

These interpretations imply that enclosures were of a lesser social status or of a lesser social function than hillforts, although this has subsequently been argued against. Some areas lack hillforts entirely, these are visible in Figure 1. The absence of hillforts within these areas implies that hillforts per se did not have a fundamental role in social organisation or social activities that could not be undertaken elsewhere. This is supported by both Harding and Hill who noted that there is no evidence for any specialist activities having taken place in hillforts which did not in enclosures (Hill 1996, 99; Harding 2004, 295). McOmish has suggested that the difference between hillforts and enclosures was that the latter had a greater longevity of use and were less disrupted than the former (1989, 108-109). Enclosures have been more disturbed by modern agricultural practices as they tend to be located in lowland areas where this activity is at its highest, whereas the majority of hillforts are located in uplands and are less susceptible to damage.

Based on morphology he produced eleven site types: • Type I – single enclosure site of 2-12ha, with single banks and ditches and a simple gap or inturned entrance • Type II – single enclosure site with stronger defences than Type I (size and/or 2 banks and ditches), inturned or entrance cut through defences • Type III – single enclosure site up to 20ha, strong multivallate defences and elaborate entrance • Type IV – Very large singe enclosure 23ha+, univallate and multivallate • Type V – small single enclosure site (less than 2ha), simple defences, simple entrance and many sites are circular in plan • Type VI – small single enclosure site, multivallate (up to 4 banks and ditches) • Type VII – Coastal promontory forts • Type VIII – Small multiple enclosure site, simple entrance • Type IX – Very large multiple enclosure site, simple entrance • Type X – Standard size multiple enclosure site, with univallate/multivallate defences, entrances same as I and II

Bradley argued for a continuum of enclosure which meant that small enclosures could still be interpreted as hillforts (2007, 247). The idea of a continuum of enclosure was maintained by Wigley who saw indications for the evolution of enclosure techniques from the simplest enclosure to the more complex and impressive hillfort (2007). These interpretations were related to an increase in the investigation of site morphologies. Typologies and classification systems In the Nineteenth Century Anderson assigned very broad typologies to Scottish hillforts; these were based upon whether or not they were of stone or earth construction (1883). However, hillforts came to be increasingly classed on a cultural basis. For example, the map of Southern Iron Age Britain was based upon geographical location and 4

Setting the Scene

Figure 1 Distribution density map of Hillforts in Britain (©Atlas of Hillforts in Britain and Ireland Project 2018; Contains OS data © Crown copyright and database right (2018))

5

A GIS-based Analysis of Hillfort Location and Morphology • • • • • • •

• Type XI – Multiple enclosure sites with enclosures physical separate RCHAMS in 1915 and Forde-Johnston based their typologies upon physical evidence as opposed to enforcing an interpretive classification. Forde-Johnston also classified the regions where the site types occurred, here he defined Type I- IV and Type X-XI as the Wessex tradition and the remainder the Western tradition (Ibid.). This demonstrates that the concept of ‘culture’ and cultural groups still had an influence within archaeological research even when the focus of the work had changed.

Unenclosed Platform Settlements Palisaded Works Homesteads and Settlements with Timber Houses Forts Dun Settlements with Stone Houses, and Field-systems. Scooped homesteads and scooped settlements

The application of social interpretation upon typological systems continued with the 1994 RCAHMS survey of south-east Perth (1994). In this instance the interpretive impact was not to the same extent as in other examples as the majority of the classes were based upon physical form, the only interpretive category was ‘fortifications’:

In Wales and the Marches, the distribution map of hillforts and defended enclosures emphasised site size. For example, Simpson defined classes as over 15 acres, 3-15 acres and 3 acres (1964). Similarly, Hogg mapped the hillforts of south-west Britain (including Wales), south-east Britain and northern Britain (including Scotland) based upon size but also using their vallation type (1975).

• • • • • • • •

Defining hillfort typologies based upon banks and ditches has been prominent across Ireland from an early date. Westropp’s study of Irish forts initially focused upon their architectural features although he later arranged them into regions, subsequently he arranged them into types (1896/1901). Looking at Westropp’s site types, ringforts seems to be the only class which could include hillforts, he divided them by size; ‘typical’ and ‘large’ sites (Ibid.). His other classes of site included ‘the walled islands’ which were marsh and lake forts, and rectilinear forts which were later in date than the curvilinear walled structures of the Bronze Age. According to Westropp, the second most important site type is the cliff/promontory fort whose defence relied upon cliffs or slopes. His final class of site was motes (both simple and complex) (Ibid.). Much later, Raftery compiled a classification system that was based upon the character of enclosing earthworks (1972). This included 3 classes, class 1 comprised sites with a single line of defence; class 2 had two or more lines, and class 3 were inland promontory forts (Ibid.).

Unenclosed circular buildings Enclosed crescents Palisaded sites Fortifications Enclosures Interrupted ring-ditches Souterrains Cropmarks of rectangular buildings

However, the concept of typologies was not adopted or agreed with by every scholar. Individually hillforts do not have a uniform morphology particularly the number of banks and ditches, for example Pen y Bannau in Ceredigion is univallate on all sides apart from its northeastern side which is multivallate. Consequently, the distribution maps that are based upon vallation sequence are a very broad generalisation of hillfort typologies that do not reflect reality. This was highlighted by Bedwin who stated that the ‘blanket term ‘hillfort’, is probably more of a hindrance than a help’ (1984, 47). Harding also argued that classification systems imply regularity in design and function that may never have been intended (2012, 14). Recent views typified by Driver have moved away from typologies and have examined the variability of hillfort morphology in Wales on a site by site basis (2005a; 2005b; 2007; 2013). Driver’s views are explored and expanded upon later in this book. However, as a generalisation, Driver examined hillforts in relation to the topography by assessing whether or not the hillfort morphology adhered to the topography, or whether a design was enforced upon the landscape. At Castell Grogwynion, for example, Driver found that in order to implement a straight façade the northern terrace had to be cut through the bedrock (2005b, 97; 2013, 132). To test this further, Driver examined and compared how individual sites responded to topographical situations. In the majority of cases, he found that the hillforts did not follow the principal of least effort as might be expected (Ibid. 2013, 133).

Instead of basing typology on the number of banks and ditches, Fox used their form, particularly their spacing. Here, she distinguished hillforts on the basis of whether they had closely spaced or widely spaced ramparts (1961). Like the map of southern Britain, Fox contextualised her discussion of hillfort morphologies with pottery typologies which led her to imply that the hillforts had cultural affiliations (Ibid.). This idea of cultural regionality was continued by Feachem who also explored hillfort typologies spatially through their morphology (Ibid.) Although the basis of typologies moved away from a primarily cultural one, there was still a strong sense that some had a social interpretation applied to them. For example, in the 1967 RCAHMS survey of Peeblesshire functional meaning was implied within the typological series with site types such as ‘homesteads’ (1967). The typological series comprised:

Driver also investigated the degree to which the morphological components to these sites were functional or symbolic (2005a; 2013). He found that topography could be utilised to falsify an image. For example, it was found that the impression of the banks which enclose a hillfort 6

Setting the Scene showed that the sites of this period were simple, but with few morphological and locational similarities (Ibid. 2001) whereas Middle Iron Age sites were more elaborate and their interiors were visible from outside of the sites (Ibid.). Hamilton and Manley’s study suggested that hillforts within south-east England were located to promote access to information (Ibid.). Initially they were placed to have maximum visual accessibility to the surrounding landscape, later they were positioned to be highly visible from the landscape itself (Ibid.).

could be accentuated and falsified by the slope and form of the land on which it was constructed. In some cases, the nature of enclosing banks and ditches and entrances went beyond satisfying a functional need towards a symbolic accessory. Driver’s assessment of hillfort morphologies required the hillforts to be broken down into their core architectural components (2005a; 2013). This allowed him to study the morphological variation within and between sites, he was no longer reliant upon static site typologies (Ibid. 2005a; Ibid. 2013). Breaking down sites into their core components allowed Driver to analyse patterns across his study area. This allowed him to question whether they exhibited evidence of regionality (Ibid. 2005a; Ibid. 2013).

Contextualised studies of hillfort location have moved beyond a chronological basis towards an assessment of material culture. For example, Driver investigated the directionality of movement within the landscape in relation to hillfort location and morphology by looking at the origin and distribution of material culture such as small finds (2005a). Chitty had also approached movement in a similar way in the Welsh Marches to answer questions of how and when people came into the area (1937). However, I would argue that the reliance upon interpreting object location and movement as the delineation of group areas and directions of movement is questionable. It is not known whether these objects came directly from the source to the hillforts, or if they had been curated and circulated for some time prior to their final deposition. If this was the case then Driver’s and Chitty’s directions of movement were not direct.

He also studied the variable morphology of the hillforts in relation to the dynamic landscape in which they sit. This landscape was defined as one of movement and visibility (Ibid. 2005a; Ibid. 2013). He accounted for how the morphology of a site varied across its circuit in relation to the location of routeways, topographical features and contemporary sites (Ibid. 2005a; Ibid. 2013). This enabled him to investigate what influenced site location and form, for example was it solely the terrain or were they designed to portray a particular image by disproportionately allocating resources within a particular area of the site (Ibid. 2005a; Ibid. 2007; Ibid. 2013). This approach was suggested by Cunliffe as a means of gaining a finer understanding of these sites (2006).

Driver also related his analysis of movement within the landscape to the visibility of hillfort interiors and the core components of their morphologies such as ramparts and entrances (2005a). This approach aided his investigation of image and function. The relationship between hillfort morphology, movement and topography had previously been examined by Mytum and Webster at Carn Alw (1989). Mytum also modelled the land before, during and after hillfort construction to assess variation in earthwork form and the changing view of the site on arrival and approach (1996).

Landscape analysis – hillfort location As shown above, positioning hillforts within a socioeconomic hierarchy was prominent within hillfort studies in the 1960s and 1970s. However, in recent years the physical position and act of ‘experiencing’ hillforts within their landscapes have come to be more central in hillfort studies. This humanistic approach is primarily based upon movement and visibility, and investigates how these factors may have informed social relations within a landscape setting. An early example is Avery’s analysis of the location of Cashel na Veen. Here he argues that its location afforded high visual accessibility both to and from the site (1991/1992).

Some of Driver’s key observations and ideas were introduced earlier in this chapter. The first of which was the investigation of the physical relationship between hillfort location and morphology (2005a and 2013). The investigation and relevance of this relationship resonated throughout the remainder of Driver’s key discussion, the second element of which was image (functional vs symbolic). This occurred when a hillfort portrayed an impression of strength and prowess (i.e. increased number of banks and ditches, elaborate entrance) within an area of high visibility or where there was a lot of passing traffic. This was defined as symbolic as opposed to functional because the physicality of the site’s architectural components went above and beyond their functional need. The portrayal of an image was also seen as front/rear symbolism, where although superficially some hillforts appeared to be complete, they were not morphologically homogenous. They portrayed a public front and a private interior/rear, for example when one side of a site had a

Hamilton and Manley investigated the function of hillforts in south-east England by examining the correlation between their location and morphology (2001). This was chronologically contextualised through an examination of the sites within their landscape context (Ibid.). This chronological approach was used to investigate changes in ideology and argued that the construction of hillforts was an expression of social and cultural meaning (Ibid.). They found that the hillforts which dated from the late Bronze Age and into the early Iron Age were located on the edge of topographical land units, but had long distance views. This led them to argue that these sites were not central places, but served to connect people and landscapes (Ibid.). This was enhanced by analysis of the site morphologies, which 7

A GIS-based Analysis of Hillfort Location and Morphology archaeology in the UK (Harris and Lock 1990). However, archaeologists who were closely associated with the planning process came to be increasingly aware of GIS applications used by planning authorities (Ibid.,36). GIS permits ‘much greater flexibility . . . [in] structuring . . . raw data and . . . [enables] both map-based and quantitative approaches’ (Ibid., 47). It allows one to display and analyse large quantities of both spatial and thematic data (Ibid.,47).

greater number of banks and ditches than the others. In military functionalist terms the greatest number of banks and ditches may have been placed within an area where the hillfort was more susceptible to attack. It could be argued that this would be symbolic if the banks were most numerous where the land was steeply sloping and/or highly visible from neighbouring and contemporary hillforts. Image was also discussed by Driver in relation to examples of ‘a correct path of movement’ (2013, 137). This was previously approached by Parker, Pearson and Richards at the Bronze Age enclosure of Springfield Lyons where a ‘correct’ path was found that kept refuse and cooking activities out of sight (1994). Driver found that in some cases, direct pathways into sites were obstructed by site morphologies, for example at Pen Dinas Elerch and Castell Grogwynion where bastions created such an obstruction. The act of obstructing direct access and routing people along particular pathways likely extended entry into these sites to enhance feelings of anticipation. In some cases, routing people in particular directions may have heightened their experience of coming across certain aspects of a site. These routeways may have been designed to both impress and conceal.

As GIS came to be increasingly adopted by archaeologists, it began to be integrated within quantitative analyses. In particular it was frequently used in predictive modelling (Wescott and Brandon 2000; Connolly and Lake 2006, 3435). It was also used to model movement and visibility within landscapes in relation to archaeological sites (Connolly and Lake 2006). This work harnesses GIS’ ability to store, display and analyse multiple types of data, to encourage the further interrogation of the data and results in the future. It also encourages the application of similar studies to other regions to enhance our understanding of the hillforts in Britain and Ireland. This study will create a body of data that will continue to increase over time and as new methodologies are developed they can be applied to an existing dataset. This would not have been as possible with the earlier non-GIS-based studies of visibility such as Renfrew’s analysis of visibility from tombs on Orkney (1979). The reliability of this work was based upon the drawing and mapping capabilities of Renfrew, and as Fleming highlighted, such studies are often very difficult to go back to and investigate due to human drawing error (1999).

These earlier investigations of hillfort location and morphology are a fundamental influence on the approach that is adopted within this study. The following section details how past approaches will be enhanced and moved forward. The enhancement of earlier approaches to hillforts within the landscape As discussed, recent studies have investigated the morphology and location of hillforts to tackle issues relating to social organisation and identity. Hillforts are not a uniform entity that can be explained through typologies (Driver 2005a; Driver 2007; Driver 2013). The current focus of questioning hillfort typologies is primarily based upon non-GIS, topographical and fieldwork analysis that are generally applied to individual areas. To develop upon this approach, and to concur with the extent of the Atlas of Hillforts in Britain and Ireland project, this study examines a series of areas which are spread across the project area. The large scale of this work meant that the interpretation of the data is on multiple levels, beginning with individual sites, then moving on to test areas; these are then compared and contrasted across regions of Britain and Ireland. The multi-scale analysis and interpretation of this study enables a broader understanding of hillforts which has never been possible before due to the constraints of small study areas and the barriers of modern administrative boundaries.

To avoid issues of human drawing error and to manage large quantities of data, this study’s analysis is primarily GIS-based. However, the results from this analysis are qualitatively enhanced by field visits and site photography. This demonstrates how the visual and physical prominence of each site is affected by movement and distance by taking photographs of the hillforts from varying distances and directions. The combination of GIS-based analysis and fieldwork has been applied to other landscape projects, for example, Arbour used GIS-based analysis both pre- and post-fieldwork to test the reliability of the field work and the accuracy of viewshed calculations (2011). Similarly, within the Lismore Landscape Project a combination of photography and viewshed analysis was used to investigate the visibility from a site (Redhouse, Anderson et al. 2002). This creates a realistic visualisation and basis for analysis and interpretation because these photographs are taken within a real environment with real weather and real vegetation which can all effect visibility.

The large scale of this work is enhanced and enabled by its predominantly GIS-based approach as this effectively analyses large volumes of data. Since the early 1990’s, GIS has been used within archaeology. GIS is a ‘data management tool and . . . a methodology in its own right’ (Chapman 2006, 9). It was slow to diffuse within

Hamilton and Manley assessed visibility both to and from hillforts, unlike many visibility studies, they explored directions of visibility and depicted it in pie-charts (2001). This depiction is not as clear as it would be by using GIS such as ArcGIS viewshed analysis. Pie-charts 8

Setting the Scene investigate whether correct pathways, as defined by both slope and visibility, existed at these sites and whether they correlate with entrance-ways or other particular morphological components such as the most extensively enclosed area.

do not inform the reader of the extent of visibility, neither can they determine which areas are visible, for example if a particular hill is visible or not. This is all possible when using GIS-based techniques such as Viewshed Analysis which is now often reported (Field and Smith 2008; Dorling and Wigley 2012). However, the use of Viewshed Analysis within these papers is based upon binary viewsheds which simply depict whether something is visible or not. As it will be shown later in this book, visibility is not definitive and it is affected by distance, atmospheric and environmental conditions.

The analytical process within the investigation of ‘correct pathways’ is a two-stage process and is explained in more detail in the next chapter. However, firstly, the methodology is applied to a group of hillforts within Driver’s study area to examine how effective his non-GIS-based conclusions were compared to those reached from this GIS-based analysis. This also aims to question Driver’s concept of a correct pathway. For example, it investigates whether the non-extant pathways (slope-based or visual pathways) correlated with a hillfort’s entrances or earthworks. This approach was inspired by the work of Lee and Stucky, and Lock and colleagues who saw visibility as a highly influential factor in pathways (Lee and Stucky 1998; Lock et al. 2014). This was undertaken by integrating visibility with movement through Cost Surface Analysis. This approach has not been applied to hillforts and it shows great potential for the investigation of movement and visibility in relation to hillfort morphology, at a scale that has never been used before, i.e. the human scale.

Moves in this direction have been made by Ruestes as she believed that visibility and distance affected and informed socio-economic relations (2008). Instead of relying upon the limitations of a binary viewshed, bands of visibility as defined by the effect of distance upon visual clarity were applied to the results of the viewshed analysis (Ibid.). The viewshed analysis results were then investigated to examine how the site’s ability to have visual control varied although she failed to clarify how one could gain such control (Ibid.). Whilst this study uses GIS-based analysis to question the archaeological record, it also questions the effectiveness of using this software as a tool to answer these questions. The questions under consideration are Driver’s key conclusions of the relationship between topography and morphology, image and correct pathways (2005a and 2013). The first of which is assessed with the analysis of LiDAR imagery, aerial photographs, site plans and DTMs. This analysis helps to examine the physical relationship between the topography and the morphology of the hillforts, to assess how the latter adheres to the former. The degree to which the form of the land is utilised within the construction of the hillfort is also examined to identify whether or not the form of the landscape in conjunction to the morphology of the hillfort was used to manipulate images of the hillfort.

Individually, visibility and access analysis have been factored into site analysis to varying degrees for some time. For example, as described above, Hamilton and Manley argued that through analysing the chronological variation of the visibility of hillforts one could understand their changing role (1997; 2001). However, they failed to take into consideration that with distance visual clarity decays. The work presented here incorporates visual decay through distance by using bands of visibility. Such banded visibility was used within Ruestes’ analyses of Iberian hillforts (2008), as described above. The difference here is that the visibility of the hillforts from the bands as well as the visibility from the hillforts is used to achieve a fuller understanding of visual variability within the wider landscape context.

The concept of image is developed further within this study, just as within Driver’s work, the idea of utilitarian vs symbolic is tackled. In particular, the spatial allocation of resources such as material, time and people in relation to the hillfort morphology is investigated. In some cases, emphasis has been put upon one side, for example the side which faces an area of a large degree of human traffic or where it is highly visible from the surrounding landscape. This element of hillfort morphology is assessed through a combination of cost surface, viewshed and fieldwork analyses. This enables the examination of the access to and visual qualities of the hillfort alongside an analysis of morphology.

This analysis of visibility enables one to question the degree to which the hillforts were integrated into the wider landscape, for example how a site ‘affords’ the act of being seen (Gibson 1979; Wheatley and Gillings 2000). This is manifested through ‘strategies of visibility’ such as monument construction’ (Wheatley and Gillings 2000) and it is these which are under investigation within this book. Strategies of visibility bring hillforts into being within a person’s world (Ingold 2000, 21). However, by integrating movement with visibility through Cost Surface Analysis, this study acknowledges the importance of movement within peoples’ engagement with the landscape (Ingold 2000, 55; Ingold 2011).

Movement is also a key feature within Driver’s third element, which is ‘correct pathways’ (2005a and 2013). Here, entrance morphology and the positioning of other features such as outworks or bastions affected movement to and through the hillfort. Although Driver based his correct pathways upon hillfort morphology, they could also be based on a least cost pathway. This study aims to

People ‘know as they go, as they journey through the world along paths of travel’ (Ingold 2011, 154). Consequently, by investigating the visual qualities of pathways this study has the opportunity to identify areas where the landscape 9

A GIS-based Analysis of Hillfort Location and Morphology assessment of their wider landscape context, the method by which this is undertaken is also detailed in Chapter 2.

afforded people the opportunity to obtain visual information about their surroundings. The process by which this study investigates both the strategies of visibility, and knowledge acquisition through wayfaring (Ingold 2011) is explained in detail within Chapter 2. However, as Barrett suggested, events and activities such as site construction all occur as a result of peoples’ ability to interpret situations with their understanding of the past and the present (1999, 24). Consequently, this study assesses the evidence for hillfort morphology and location through a contextualised

Chapter 1 highlighted the theoretical underpinnings and influences to the approach within this book which set the scene for the study. Chapter 2 subsequently develops upon this introduction to introduce the methodology used as well as highlighting the fundamental research questions asked and how the methodology is used to answer them.

10

2 Methodology and Research Questions hillforts, its multiple-area approach has produced a large quantity of data that can only be effectively managed and analysed within a GIS.

The principal focus of this research is to test how effective Geographical Information Systems (GIS) are to investigate and compare the various influences on hillfort location and morphology. The key influences focused upon are topography, visibility and physical accessibility which are investigated through the application of Cost Surface and Viewshed Analysis. The intention is to explore and assess the three key elements of Driver’s study of hillforts in Ceredigion (2005a; 2013) in which he identified and analysed:

Initial methodological stages: test area selection and the acquisition of a DTM Test area sampling During this project’s developmental stages multiple means of deciding upon test areas were investigated. The first of which was to select areas purely based upon topography as topography is undeniably a highly influential factor in the location of hillforts. Consequently, by identifying topographically variable and similar locations the writer is able to examine how these locations influenced site morphologies. This method did however encounter problems from the outset, the most significant being data acquisition. Although digital topographical data is available for Scotland, Wales and England but not for Ireland, it is in different formats. Topography is also locally highly variable, therefore classifying areas by topography across all countries would be complex and ineffective so this method was not considered suitable within the time frame available.

• the use of image • correct pathways • the physical relationship between hillfort morphology and landscape topography The significance of the application of GIS-based analysis to hillfort location and morphology To outline, reinforce, and test these three core elements of Driver’s study, this work applies the GIS-based methodology, developed here, to a subset of the hillforts which were within Driver’s study area (Chapter 3). By testing Driver’s conclusions against a GIS-based analysis it will be shown whether if Driver utilised GIS, he would have drawn different conclusions or whether GIS supports and enhances his observations. GIS is able to assess to what extent distance and direction affected the visibility of different architectural aspects of the sites and how that related to their image. Through the use of GIS, the concept of a correct pathway can also be analysed further. For example, one can investigate to what extent these extant pathways such as entranceways and known routeways correspond with visual and slope-based pathways. These non-extant visual and slope-based pathways can also be investigated further using GIS, by investigating their correlation with the architectural components of sites.

The process of random sampling was also investigated, here a homogenous grid would be placed over the distribution map of the total hillfort population of Britain and Ireland and a process of random selection would then be used to select test areas. This process was not adopted as the aim of a random sample is that it is ‘representative’ of a total population with ‘the part . . .stand[ing] for, or represent[ing], the whole’ (Orton 2000, 14). The aim of this research is not to use the test areas to predict where architectural features would occur at hillforts within different regions. Rather the aim is to investigate whether movement and/or visibility influenced the location and morphology of the test area hillforts alone and to assess differences between areas. It also investigates whether a GIS-based methodology is capable of answering qualitative questions about the location and morphology of individual hillforts.

The significance of a multi-regional approach to hillfort location and morphology Whilst there has not been a GIS-based analysis of hillfort location and morphology, Chapter 1 highlighted that there has also not been a multi-regional approach to hillfort location and morphology in recent times. Both Driver, and Hamilton and Manley’s individual studies focused upon small regions and there was little, if any, in-depth comparison made with other areas (Driver 2005a; Driver 2013; Hamilton and Manley 1997; Hamilton and Manley 2001). It is impossible to effectively assess the morphology and location of every hillfort within Britain and Ireland consequently it was decided that this study would be based upon a series of widely spaced test areas. Even though this study only uses a small subset of Britain and Ireland’s

It was decided that the process of test area selection would be multi-phase. This study’s primary geographic focus is Britain, consequently it is important to select from the entire hillfort population and to sample areas with differing degrees of hillfort activity, as represented by hillfort density. Producing the density map required the collation of point data for all of the hillforts across Britain and Ireland into one shapefile which was then input into the point density tool in ArcGIS. Areas of differing densities were subsequently visually selected from the distribution map. 11

A GIS-based Analysis of Hillfort Location and Morphology obtained and processed thus substantially increasing the analysis time. Consequently, subsets of the test areas (apart from Aberdeenshire) were analysed with a LiDAR based DTM to investigate the differing effect that DTM resolution had upon the results of the GIS-based analysis. These DTMs came in ASCII files in various resolutions from 50cm to 2m. This data was straight off the shelf from the Environment Agency and Natural Resources Wales, further processing was not undertaken. Further processing was not considered necessary as the LiDAR data was studied alongside aerial photographs, site plans and site visits. The combination of multiple data types counteracts the effect that any defects in the LiDAR data would have on subsequent interpretations. The difference in the results of large scale GIS-based analysis between the two DTMs was minimal thus the study was able to proceed with confidence in the use of the DIGIMAP derived DTM for analysis.

Once areas of different densities were selected, the underlying topography was also examined. To effectively test the methodology, it was important to have a series of topographically different areas, this variation could be both within and between test areas. The topographical characteristics of the test areas were qualitatively investigated through an examination of the Digital Terrain Models (DTM), and an assessment of the distribution of watercourses and hillforts within these regions. As a final step, closer examination of these topographically and distributionally variable test areas required further inspections of their hillforts themselves specifically their morphologies to ensure that the groups of sites chosen were morphologically variable. This would ensure that the methodology was tested to its maximum. The acquisition of Digital Terrain Models

Despite this, it was decided that throughout the work the results of the LiDAR based analysis would be displayed alongside those from the Ordnance Survey DTM based friction surfaces. This shows the similarities and differences between the results to enable a fully informed interpretation of them. LiDAR was the only friction surface which was used to investigate the differing degrees of visibility to the architectural components of the hillforts. Whilst the cell size of the PROFILE DTM was sufficient for the large-scale GIS-based analysis, it was too great to depict the finer detail for the analysis of individual architectural components to the sites. However, due to the large size of the Ceredigion test area, LiDAR was only used to investigate the contradicting results of Driver’s observations with the results of GIS-based viewshed analysis, specifically in relation to the intervisibility between Gaer Fawr and Pen Dinas.

A GIS-based study’s foundations lie within the acquisition and effective use of a DTM. The DTMs for the test areas were retrieved from EDINA Digimap (in dxf format). This is a PROFILE DTM which comprises 10m grids of height points derived from the PROFILE contour data (Ordnance Survey 2012). For PROFILE contours, vertical values are recorded every 5m generally and every 10m in mountain and moorland areas (Ibid.). To counteract small residual errors within the construction of the DTM, the common edge between the DXF tiles is blended (Ibid.). However, Ordnance Survey note that if the data was provided at different times blending would not have taken place, and there may be small differences in the height along the adjoining edges of tiles (Ibid). For this work some data was retrieved at different times, and consequently there may be very slight differences within height data, however, having assessed the DTMs, the effect was seen to be negligible. It was however noted that in the Aberdeenshire test area a portion of the DTM is poorly blended as the outline of the tiles are relatively prominent although this was not caused by requesting data on different days. Even so it is not likely that this would have greatly influenced the results of the GISbased analysis, as this erroneous area is predominantly a central lowland region between upland regions.

Approaches Site morphology in relation to topography A key aspect of Driver’s work was his examination of how hillfort morphologies relate to their topographical position (2005a; 2013). This allowed his study to investigate whether topography aided the construction of the site. It also meant that he could investigate how the morphological variation of a site, for example the number of banks and ditches which enclose it, relate to the land that it is situated on.

Ideally, the DTM would have been based upon LiDAR data retrieved from the Environment Agency as this is at a much finer resolution (Environment Agency 2015). However, no data was available for the Aberdeenshire test area and in some cases those areas that had LiDAR had incomplete datasets which would have a detrimental effect on the results of GISbased Cost Surface Analysis. Similarly, higher resolution DTMs often run the risk of having an increased number of artefacts. These are introduced through high intensity processing and would influence the results of GIS-based analysis, consequently it would have to be used with caution.

Method of investigation To gain a detailed understanding of site morphologies in relation to the topography, the PROFILE DTM was not sufficient enough consequently the LiDAR data was used. This LiDAR was processed into a hillshade model. This process has been identified by many to greatly effect data interpretation (Kokalj et al. 2013; Opitz 2013; Opitz and Cowley 2013). Hillshade modelling helps to gain a better understanding of the morphology of sites and to identify specific aspects, particularly:

Another consideration is that the test areas within this study are large, spanning beyond 9km around any one site. This would have required a large quantity of LiDAR data to be 12

Methodology and Research Questions site) (Figure 3). Its effect on the smaller site was greater (Figure 2). However, the visual magnitude of this site was very low prior to the addition of height to the bank, consequently the changes in the visibility of areas of this site was from low visibility to not visible.

• Form and extent of enclosing earthworks • Presence of quarry scoops • Location of interior features such as barrows, enclosures/annexes The hillshade models were created by converting the ASCII files to rasters, these were then mosaicked into a single raster within ArcMAP. Various multi-directional hillshade models were created from these rasters which enhanced the visibility of archaeological features.

Visibility Earlier methods of investigation Visibility has featured in hillfort studies for some time with Driver (2005a and 2013) and Hamilton and Manley (2001 and 1997) undertaking non-GIS-based visibility analysis. Visibility studies are believed by many to enable researchers to develop a better understanding of past social networks. For example, Jerpåsen acknowledged the potential effectivity of applying GIS-based analysis to applications of the Archaeological Landscape Analysis by Visual Methods methodology (2009). This involves investigating landscapes as ‘Landscape Rooms’ and it has the potential to provide a methodology to phenomenological research (Ibid.). Brughmans, Keay and Earle used a combination of site visits, line of site calculations and exponential random graph modelling to investigate visibility networks in Southern Spain during the Iron Age and Roman periods (2015). This study represented sites through single observation points which they openly stated was problematic, however the nature of these sites meant that it was problematic to define their true extent as it was not known. Brughmans, Keay and Earle also emphasised the importance of combining GISbased analysis with field visits because the paucity of their DEM meant that there were instances where the GIS was not picking up lines of sight that were visible in the field (Ibid.).

Once a clear understanding of the morphology of the individual hillforts was established, through the construction of the hillshade models and the assessment of site plans, the core architectural components of sites were marked out by creating a polyline along the peaks of the banks. This was subsequently split into points which were placed every metre along the top of the enclosing works. Whilst the original height of these banks is an unknown, due to a lack of excavation it is also not known whether they had a palisade. Any additional height would have affected visibility both into and out of the site. The primary focus of this study is the visibility of hillforts and their architectural components consequently tests were undertaken to investigate how the height of the inner bank could affect the visibility of a site’s interior. The inner bank was used as a test case as this is usually the biggest, and this test was applied to two sites of varying character. The first of which was Hardings Down North Enclosure, this is a hillslope enclosure that encloses 0.2ha (Figure 2). The visibility of this site was assessed from The Bulwark. To test whether site size influences the effect of the height of a site’s enclosing works has upon the visibility of their interior, the second test was applied to Scratchbury. Scratchbury encloses 17ha (Figure 3) and the visibility of this site was assessed from its neighbour Cley Hill.

Further studies of past visual networks include Marsh and Schreiber’s investigation of the positioning of imperial sites within the Wari Empire (2015). This investigated the visibility of the landscape from known sites compared to from random locations (Ibid.). They challenged the null hypothesis by identifying that imperial sites were positioned within locations that have a greater degree of visibility to the surrounding landscape than random locations (Ibid.). Visibility investigations of known sites compared to random locations were also employed in relation to chullpas, which were found to be located within highly visible and highly elevated locations (Bongers et al 2012). Additionally, Mitcham undertook GIS-based analysis to investigate how defensible Hampshire hillforts actually were (Ibid.). This investigation was based upon the belief that a site could have a defensive role if:

The cell size of the PROFILE DTM was too large for it to be used effectively as a DTM; consequently, this investigation relied upon LiDAR data. Although height was added to the compacted bank surface, the width was determined by the resolution of the DTM. Heights of 1m, 1.5m and 5m were added to the known height of the bank. 1m was chosen as a minimal starter height. 1.5m was chosen as it was postulated that the height of any additional breastwork would not be much higher than an individual’s chest height (Ian Ralston pers comm). Consequently, as the study uses a height of 1.7m for an individual, the chest height is 1.5m. 5m was also chosen as this took into account an additional, pre-erosion bank height and an additional palisade.

• It was on the steepest slopes • It had good views of the immediate surroundings • Its interior was not visible

Figures 2 and 3 illustrate the variable affect that the addition of height to the inner bank of these enclosures has upon the visibility of their interior and shows a generally limited effect. Its primary influence on the larger site of Scratchbury was upon the visual magnitude of the visible areas (the number of observer points which can see the

By investigating the results of viewshed analysis and analysing slope maps, Mitcham identified that not all of her study hillforts were defensible (2002). 13

A GIS-based Analysis of Hillfort Location and Morphology

Figure 2 Result of investigations into the effect that inner bank height has upon the visibility of Hardings Down North enclosure from The Bulwark (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

14

Methodology and Research Questions

Figure 3 Result of investigations into the effect that inner bank height has upon the visibility of Scratchbury from Cley Hill (© Environment Agency copyright and/or database right 2015. All rights reserved.)

15

A GIS-based Analysis of Hillfort Location and Morphology landscape was also analysed to establish where the hillforts were visible from. This was calculated both through a visual dominance grid and from undertaking viewgrids (explained below) from the buffers (1km, 3km, 6km and 9km) which surround the site.

However, visual relationships are affected by distance and this has been factored into multiple GIS-based studies. For example, Carme Ruestes factored in distance into her investigations of visual control from sites within Iberia through undertaking banded viewshed analysis (2008). Similarly, Wheatley and Gillings emphasised the importance of undertaking viewsheds both to and from sites as opposed to wrongly assuming visual reciprocity by only undertaking viewsheds from sites (2000, 7-8). They argued that as the viewing distance in relation to viewer height increases, the visual reciprocity between views to and from a point increases (Ibid. 2000, 8). This was undertaken by overlaying the results of viewshed results with a binary Euclidean distance calculation (Ibid. 2000). Llobera also incorporated distance into viewshed analysis by investigating whether visibility within different topographical areas differed when being in the foreground, middle ground or background (2003). Similarly, by examining views of the Ridgeway hillforts from short and long distances, Bell and Lock investigated whether they were positioned to guide people along points of the route (2000).

Through the calculation of visual magnitude, defined as the degree of visibility, this study examines how many times a particular part of a site was seen and from where. This is similar to the work of Llobera who calculated the visual magnitude of areas to see whether or not topography and location, for example valley side, upland/lowland or ridge, afforded homogeneity or heterogeneity to visualscapes (2003). Directionality is also influential within visibility studies. Through his qualitative analysis, Driver argued that there was evidence for morphological directionality which was manifested in the portrayal of particular images within a particular direction at hillforts in Ceredigion (2005a; 2013). With regards to GIS-based studies, Wheatley and Gillings argued that viewshed analysis could be enhanced by taking into account the directionality of visibility by overlaying the aspect raster onto the terrain surface (2000). Within this book, directionality is explored through a GIS examination of the differing values of visual magnitude of different morphological aspects of sites. This allows an investigation of the differing visual qualities of hillforts and their morphological components. It establishes whether there is quantifiable evidence for the portrayal of a particular image within a particular direction, for example it allows one to see if the strongest side of a site is also its most visible.

The effect of distance on visibility, particularly clarity of vision has been influential within viewshed analysis since the recognition of the work of Higuchi (1983). Higuchi argued that short, middle and long-distance views affected visual clarity, this was based upon the visibility of trees. Short distance was interpreted to be sixty times the size of the object and you could achieve a steady gaze upon an individual tree (Ibid.,14). Mid- and long distance varied with the type of tree with the transition from mid to long distance visibility occurring at the point where individual trees could not be distinguished (Ibid.,15). His work also integrated the effect of topography on visibility; this is highly applicable to the study of the visibility of hillforts (Ibid.). Much later, Ogburn investigated the visibility of cultural objects within landscapes (2006). This methodology was an adaptation of Fisher’s fuzzy viewsheds, here Ogburn took into account target size alongside visual acuity (Ibid.).

Whilst GIS-based analysis is the basis for investigations of visual directionality it is also combined with field visit observations. This involves analysing the degree to which the hillforts were visible from the surrounding landscape. Here photographs and field notes were taken from varying distances and directions to illustrate how the image and clarity of the hillfort differs through movement, distance and direction. Similarly, the visibility of the surrounding landscape from the hillforts was also assessed by taking photographs from the cardinal points of the sites.

Current focus of the investigation The current work develops upon these earlier GIS-based approaches to visibility, with the aim of answering the following questions:

The basis of viewshed calculations The viewshed analysis is based upon Visual Dominance (VD) and View Grids (VG) (Lock et al. 2014). Visual Dominance calculates the view to a point and the resultant viewshed depicts the areas that a point is visible from (Figure 4). A view grid calculates the view from a point with the resultant viewshed depicting the areas that are visible from it (Figure 4). Both VD and VG are calculated using the Viewshed Analysis tool in ArcMAP, with each using modified values of OFFSETA and OFFSETB. OFFSET is the vertical distance above the observation point for example the height of the observer (Esri 2012). OFFSETA is the vertical offset of the height of the observer, whilst OFFSETB is the vertical offset of the height of the target.

• What was the degree of visibility from the hillfort? Were there long distance or short distance views? • How visible was the interior of the hillforts from other hillforts? Was one area of the hillfort more visible than others? How does that relate to the morphology of the hillfort? Method of investigation Viewshed analysis was undertaken from the hillforts to analyse how visible the surrounding landscape was from within them. The visibility of the hillforts from the 16

Methodology and Research Questions View Grid

Visual Dominance

View Grid (Looking out of the hillfort):

View Dominance (Looking into the hillfort):

OFFSETA- 1.7

OFFSETA- 0

OFFSETB- 0

OFFSETB- 1.7

Figure 4 Schematic explanation of VG compared to VD

encouraged the computation of viewsheds across large areas (2013). Consequently, instead of investigating the visibility of the hillforts via a Visual Dominance calculation (VD) or from the hillforts when calculating a View Grid (VG) towards a single point, this work opts for a grid of points. This grid was placed over the interior of each hillfort using the Fishnet tool within ArcMap. It comprised 49 points (7 by 7). A 49-point grid was chosen having experimented with multiple grids of different sizes (25, 49, and 100 (Figure 5)). This experiment was based upon a Visual Dominance calculation and indicated that the grid size only affected the number of points in the hillfort that were visible but not from where they were visible.

This works on the basis of visual reciprocity. A height of 1.7m is used for the observer value because osteological evidence from the Iron Age has indicated that this was the average height of a man (Chapman 2006, 85). For each series of viewshed calculations, corrections were applied to allow for earth curvature. Kormann and Lock illustrated the effect of applying a curvature with refraction algorithm to a DEM for viewshed analysis (2013.). They found that earth curvature, coupled with refraction, had its greatest affect beyond 9km where the percentage change of using a standard viewshed, compared to applying curvature and refraction, was 16% (Ibid.). As this work primarily works upon distances within 9km, the decision was made not to apply this algorithm and that the standard ArcGIS routine for curvature refraction would suffice.

The irregular shape of some hillforts compared to the square-shaped point grids meant that in some instances the grids extended slightly beyond the hillfort perimeter. This does not hinder the nature of the results. Although the hillfort interior was defined by the enclosing works, the land immediately beyond these earthworks was still important as activity associated with the hillfort would have taken place there. Consequently, it is also important to understand the visual relationships within these areas.

Visual distances Distance was factored into the current research by creating a series of buffers around the sites, based upon Ruestes’ work on Iberian hillforts (2008). She utilised bands of visibility that were set at 1km, 3km and 6km arguing that an increase in distance led to a decrease in visual clarity. At up to 1km, Ruestes saw visual clarity to be perfect, whilst at over 6km although there was still visibility it was not always possible to determine what was being looked at (Ibid., 368). Based on this, buffers of 1km, 3km, 6km and 9km from the centre of the hillfort enclosure were chosen. An extra 9km buffer was constructed to create a known cap on the visual distances. Points were placed every 500m along the circumference of these buffers, viewsheds were then generated from each of them.

The study also develops the multiple-point viewshed approach by investigating the visibility of sites from their buffer radii. This involved placing points every 500m along the buffers and undertaking viewsheds (VG- Looking out) from each point. The visual representation and management of viewshed results These viewshed analyses created a large amount of data. To effectively analyse and examine the visual magnitude of these sites, the means by which this data was displayed had to be modified using the Classification tool within ArcMAP. This excluded values of 0 and organised the data into four classes which were defined by a quantile interval. The cumulative viewshed was therefore organised into percentiles, (1-25 low visibility 75-100 high visibility)

Multiple point viewsheds To rigorously examine the visual relationships within a landscape, this study moves away from single point to point viewsheds towards a multiple-point approach. This was inspired by the work of Tabik and colleagues who questioned the reliability of single point viewsheds and 17

A GIS-based Analysis of Hillfort Location and Morphology

Figure 5 Comparison of the effectiveness of 25, 49 and 100 grid point viewsheds, depicting the visual magnitude of Moel Arthur (Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

18

Methodology and Research Questions cost surface analysis between focal sites and sites with artefactual and faunal evidence that were indicative of later prehistoric activity, alongside the computation of ‘control paths’ (2012). The ‘control paths’ were designed to travel between the focal sites and every 5km directional zone within 40km (Ibid.). This enabled them to gain a general picture of potential areas of movement which were not reliant upon the limitations of the archaeological record (Ibid.).

which enabled easy comparison of values in visual magnitude within and between sites. As this study is highly visual, it was recognised that the effective use of colour as a means of communication was paramount. Colour can enable a reader to intuitively read images (i.e. hot reds are high values and cool blues lower values (Figure 5). This concept is based upon the work of Bertin who argued that the use of the three primary colours enables one to ‘differentiate meaningful values that the eye would only be able to distinguish with difficulty otherwise’ (1981, 223).

Güimil-Fariña and Parcero-Oubiña did not utilise the ‘from everywhere to everywhere’ model, instead they investigated why roads were positioned where they were (2015). They examined the coincidence of least cost pathways with known Roman roads and milestones with the aim to support their hypothesis. This hypothesis was based upon the idea that Roman roads would have been positioned where movement required the least amount of effort (Ibid.). Similarly, Symonds and Ling attempted to develop an understanding of social practice within a landscape through an analysis of artefact distribution (2002). They investigated travel time along Roman roads and rivers between places of pottery production and consumption in Lincolnshire (Ibid.). The approaches detailed within this brief synopsis highlight the importance of relating GIS generated pathways with known sites, topographical features and areas of activity which was a fundamental influence on this methodology, as detailed below.

Movement Earlier methods of investigation Movement to and from hillforts has also factored into landscape-based hillfort studies, Driver, for example, experientially analysed the accessibility of hillforts from the surrounding landscape and their visibility from routeways (2005a and Driver 2013). As with earlier visibility investigations, earlier GIS-based investigations of movement have been used to investigate cultural interaction and social networks. However, the approaches of these studies have varied in a number of ways. The degree to which landscapes afforded movement has been investigated. Cost Surface Analysis was used to investigate the terrain’s influence upon movement along the ridgeway in Oxfordshire (Bell and Lock 2000). Llobera and others studied the physical accessibility of hillforts by investigating topographically afforded mobility networks (2011). Similarly, whilst Howey did not conduct cost surface analysis she examined the permeability of a landscape in North Michigan (2015). This investigated how a landscape enabled cultural processes by looking at it as a continuous space as opposed to a series of points through permeability modelling.

Current focus of the investigation Cost Surface Analysis is used here at multiple levels to answer the following questions: • How physically accessible were the hillforts? • How did the orientation of least cost paths relate to the morphology of the banks, ditches and entrances of the hillforts? The key aim of this work is to examine how the pathways enter or exit the hillforts and how that relates to the morphology of the site. This allows the investigation of how morphological variation related to physical accessibility. For example, were the entrances placed in the most accessible aspect of the site and did the most prominent and numerous banks that enclose the site face its most accessible area?

Taliaferro, Schriever and Shackley utilised slope-based cost surface analysis to investigate trade networks in Mimbres, New Mexico (2010). This study constructed least cost pathways between consumption sites and the sources of obsidian to investigate social interaction within the landscape through movement and obsidian procurement (Ibid.). Investigations of resource procurement through GIS-based analysis have also factored in travel time (McCoy et al. 2011). Hazell and Brodie undertook cost surface analysis based upon a friction surface of a slope gradient and area avoidance to investigate megalithic transportation routeways (2012).

Method of investigation The least cost pathways calculated here went between the hillforts and other hillforts, and the buffer areas. The computation of least cost pathways between the hillforts and their radii aligns the study with those such as White and Barber (2012), and Nolan and Cook (2012) who explore general patterns of movement and accessibility on a landscape scale. Cost surface analysis was based upon anisotropic cost surfaces, which take directionality of movement into account. Directionality is an important

However, GIS-based studies have increasingly investigated the accessibility of whole landscapes. For example, the ‘from everywhere to everywhere’ model was used by White and Barber to examine how the routes of least cost pathways based upon topography and land cover compared to known pre-Columbian and early colonial movement corridors (2012). Similarly, Nolan and Cook undertook 19

A GIS-based Analysis of Hillfort Location and Morphology • Earthworks – sites which are listed as a bank, ditch, reave, linear bank and ditch. It was important to include these as they were the result of collective labour and may have been influenced by, and influenced, the nearby hillforts. • Funerary – round barrows, long barrows and cists. These sites are often found within and near to hillforts and it is important to investigate this physical relationship and the influence that these earlier sites had upon the morphology and location of hillforts. • Agriculture – sites listed as fields or field systems, earthworks or cropmarks. This category is important as they were the result of periodic episodes of collective labour that was potentially influenced by hillforts or influenced hillfort location. • Occupation – this category comprised sites listed as unenclosed and enclosed settlements that were of prehistoric date. It was important for these sites to be included as they were potentially in use at the same time as hillforts. • Finds – the distribution of finds was also investigated to identify possible areas of activity.

factor within cost surface analysis as all movement has direction and moving down a 20° slope is not equivalent in cost to moving up a 20° slope (Bell and Lock 2000, 89). The simplest cost surface analysis used here was defined by slope which is considered to be movement’s most influential factor; it dictates the type of movement which may have taken place within a landscape. It also dictates whether only foot traffic or whether carts and goods could travel along particular routes. This study uses the TAN (Slope)/TAN (1°) equation (Ibid.) which is based upon ‘the ratio of the tangents of the slope angles . . . [as opposed to] using the angle of slope itself’ (Ibid., 88). This takes into account that cost does not increase linearly with an increase in slope. It also factors in that people may walk across slope tangents as opposed to directly following them (Ibid. 2000). The entire process of creating the friction surface and generating the least cost pathways was applied to both the PROFILE and LiDAR DTMs. To create the slope-based friction surface, a slope degree surface was calculated using the slope tool within ArcGIS. This output was then processed in Raster Calculator using the following equation to create Bell and Lock’s slope-based friction surface:

Correlations between the HER data and the GIS-based least cost pathways would imply that these areas were places of significance at the time that the hillforts were in use as there is evidence that people travelled through them and occupied them. By moving around and being within these areas, the landscape and the sites within them became visible and consequently came into ‘being in that person’s world’ (Ingold 2000, 21).

Tan ((“Slope_deg” x math.pi) / 180) / 0.01745 To undertake anisotropic cost surface analysis, this friction surface was applied to the Path Distance tool. This used both linear horizontal and vertical factors to calculate the Cost Distance and Back Link rasters. These were then input into the cost path tool to calculate the resultant least cost pathway.

Movement AND visibility Earlier methods of investigation

Expanding upon the use of slope-based cost surface analysis, least cost pathways were calculated from each buffer point to the hillforts, as well as between the hillforts (hillfort to hillfort movement, both to and from them). These pathways were then reclassified and summed to calculate the frequency of the pathways within areas of interest.

Investigations of the defensibility of sites have identified the importance of both visibility and accessibility. For example, Sakaguchi, Morin and Dickie used a combination of least cost pathways and viewshed analysis to explore the defensibility of large prehistoric sites in the mid-fraser region on the Canadian plateau (2010). Similarly, Hudson’s landscape scale space syntax study applied cost surface and viewshed analysis to the investigation of social organisation and settlement patterns (2012). These factors were explored by investigating the openness and accessibility of prehistoric pueblo habitation sites (Ibid.).

Whilst the primary focus of this study is at the site scale, to reinforce the significance of any site based morphological correlations with pathways, and to increase the likelihood of these pathways being ‘real’ they also had to be investigated on a landscape scale. The landscape scale analysis involved the study of the pathways in relation to the topography and activity which was contemporary with the hillforts. Such correlations also suggest that the GISbased ‘correct pathways’ also related to topographically defined ‘correct pathways’ thus increasing the likelihood that these may have been a through route at the time that the hillforts were in use.

Kirst and Brown used a combination of cost surface analysis and viewshed analysis to predict caribou migration routes and to identify suitable hunting locations (1994). Doyle, Garriston and Houston also utilised a combination of cost surface and viewshed analysis to investigate the possible effect of exchange routes on settlement locations in the Southern Maya lowlands (2012). In this case, cost surface analysis was used to identify the least effort route through the landscape and viewshed analysis was subsequently used to investigate the visibility of these routes and how that may reflect territoriality (Ibid. 2012).

Least cost pathways were also related to the distribution of known sites and areas of activity within the landscape through the analysis of HER data. HER data dating to the Prehistoric period was categorised to enable easy comparison. The categories were: 20

Methodology and Research Questions places in the landscape where approaching traffic has least visibility of the hillfort enclosures. In particular, the following questions were asked:

Chapman investigated what was believed to be a route between two enclosures on Sutton Common (2000). In particular, he investigated the changing visibility of morphological aspects to these enclosures by undertaking viewshed analysis at set points along the route. Three years later he used a similar methodology to investigate the position and morphology of Rudston ‘Cursus A’ through a combination of cumulative viewshed and cost surface analysis (2003).

• Were the most prominent architectural components placed to face where approaching traffic is most visible? • Were the most prominent architectural components placed to face where approaching traffic has least visibility of the upcoming hillfort? • How do the visual pathways relate to the entrance/s of hillforts? • Do the visual pathways coincide with any topographical and/or archaeological features within the wider landscape, which could be contemporary with hillfort activity?

Chapman’s approach, alongside others previously highlighted, illustrate that GIS-based studies have increasingly studied visibility and movement in relation to both hillfort and non-hillfort sites. However, these two factors also potentially had an influence over each other. For example, navigation through a landscape may have been influenced by what could or could not be seen, this has not been explored in relation to hillforts until now. This approach develops upon the work of Lee and Stucky who argued that visibility affected movement, they examined movement through the landscape according to varying degrees of visibility (1998; Lock et al. 2014). Visibility was used as the sole cost factor within their cost surface analysis (Lee and Stucky 1998; Lock et al. 2014.), however here it is integrated with slope. This study also does not use the exact pathways as those that were used within the earlier works. To avoid any confusion the pathways were given completely different names, these are: blind and visible pathways. The definition of these is detailed later in the chapter.

Method of investigation The initial weighting of the visual least cost pathways was to factor visibility 40% to slope 60%. However, after initial results it was decided to increase the weighting to draw out the visual factor and weight this at 60% of the cost within the friction surface. At a later stage the question was raised as to how the weighting of these friction surfaces influenced the overall route of the least cost pathways. This led to the creation of multiple friction surfaces with the visibility: slope weighting as follows: 60:40, 25:75 and 100:0. These friction surfaces were used within cost surface analysis which was based upon the PROFILE DTM. To reduce processing time but to also allow for the comparison of the results from the PROFILE DTM and the LiDAR based DTM, a LiDAR based friction surface for the 100% visual pathways was also used within the Cost Surface Analysis.

Current focus of the investigation Visibility was integrated with movement within this book to investigate how the different visually categorised pathways relate to site morphologies thus allowing a more detailed examination of the visual qualities of sites. For example, by calculating blind and visible pathways, the study is able to identify areas within the landscape where sites are most visible to approaching traffic, or to identify

The blind paths followed a route where the origin and destination were least visible from the pathway. This path is based on the sum of the Visual Dominance grids for the origin and destination (VDa + VDb) (Figure 6). Put simply the results of the cumulative viewshed influence on the cost surface analysis is as follows: • 0 – No penalty for locations which cannot see a or b. • 1 – Low penalty for locations which can see a or b. • 2 – High penalty for locations which can see a and b Visible pathways are those paths that follow a route most visible from a or b, the friction surface is based on the inverted view grids (Figure 7)

Figure 6 Calculating the blind pathway friction surface

Figure 7 Calculating the visible pathway friction surface

21

A GIS-based Analysis of Hillfort Location and Morphology inverted VD grids for the individual architectural components (Figure 8). These VD grids were constructed by undertaking viewshed analysis every 1m along the components, primarily the enclosing banks of the sites (the DTM used within this analysis was the LiDAR DTM only). For every architectural component (e.g. bank) a cumulative viewshed was constructed which formed the basis of the friction surface for the cost surface analysis. Unlike the blind and visible friction surfaces, the friction surface for the VD of the different architectural components were not integrated with a slope-based friction surface as the primary focus here was to identify where the individual components were most visible from, and not their actual physical accessibility. The cost surface calculations followed the same principle of the blind, visible and slopebased least cost pathways as linear horizontal and vertical factors were applied within the Path Distance tool.

The inverted cumulative viewshed results influence the cost surface analysis as follows: • No penalty for locations which are seen from both a and b • 1 – Low penalty for location which are seen by either a or b • 2 – High penalty for locations not seen by a or b To combine both the slope-based and visible friction surfaces they had to be individually standardised, based on: (Friction surface – min (Friction surface)) / (max (Friction surface) – (min (Friction surface)) The standardised friction surfaces were then combined and weighted, for example a 60:40 weighting for the visible pathway meant that the friction surfaces were combined as follows: (0.6*Standardised Visible (0.4*Standardised Slope)

friction

surface)

For each morphological element, a least cost pathway was created which depicts the route through the landscape from which the element is most visible (i.e. the most number of points are visible). This aids the examination of the site’s ‘image’ for example whether it portrayed a whole or fragmented image towards the surrounding landscape. If a whole image was portrayed then all of the paths for each component would follow the same route.

+

The resultant combined friction surface was then used as the friction surface for the Path Distance tool. As with the slope-based least cost pathways, linear horizontal and vertical factors were applied within this tool. The least cost pathway was then calculated using the Cost Path tool.

As with the blind and visible pathways the friction surface’s effect on the route of the pathway is that the higher the value, the higher the penalty. As this visual dominance grid was inverted, this pathway follows the route where the most points are visible as they have the lowest penalty. Although these examples were based upon single point viewsheds, the principle is still the same with multiple points.

The least cost pathways were subsequently reclassified using the Reclassify tool and were then summed using the Cell Statistics tool. This enabled a more detailed investigation of the frequency of the correlation of blind and visible pathways with features of interest including the morphological components of the hillforts, and topographical and archaeological features within the wider landscape.

The basic principle of this cost surface analysis is: • 0 – No penalty for a location which can see a point • 1 – Penalty for a location which cannot see a point

The application of cost surface analysis to the investigation of the visibility of different architectural components to hillforts

Conclusion Alongside Chapter 1, this chapter has provided both the theoretical and methodological underpinnings to this book. It has highlighted the research questions and detailed the methodological steps which were taken to answer them. Chapter 3 develops upon these first introductory chapters by examining Driver’s Ceredigion study area with GISbased analysis. This explores Driver’s key ideas of ‘image’, ‘correct pathways’ and the ‘physical relationship between morphology and topography’ more closely to cement the foundations of this research before moving on to apply the methodology to other areas. Chapter 3 also sees the application of the GIS-based methodology to some of Driver’s sites to investigate the impact that GISbased analysis has upon non-GIS-based results.

Cost Surface Analysis can also be used to investigate the visual image/s of hillforts based upon the use of the

Figure 8 Explanation of the friction surface for calculating a routeway where architectural components are most visible within the landscape.

22

3 Ceredigion Introduction

Introduction to the sites

Chapter one highlighted the underlying influence that Toby Driver’s study of the location and morphology of hillforts in Ceredigion had upon the approach which is adopted within this GIS-based analysis (Driver 2005a; Driver 2007; Driver 2013). Driver’s work and ideas structure the discussion and core foci of this piece of research. This analysis focuses upon Driver’s three principle conclusions and observations, which were:

Located in the south-eastern corner of the test area is Pen y Bannau. This site is situated on the western end of an upland plateau. The summit of the crag creates a spatial division within the centre of the enclosure. The extent of the site is largely defined by the topography and is enclosed by a singular bank (Figure 10). The bank that encloses the southern aspect of the site is believed to have been a footing for a timber palisade due to its low position (RCAHMW n.d.a). The northern defences form the site’s primary enclosure and are its most impressive morphological component. At the northern terminal of the crag, the ramparts turn in to form the main gateway which is flanked by two further ramparts (Ibid). An additional entrance is located on the western side of the southern half of the fort; however, access to this is from a steep slope (Ibid.).

• The interplay between hillfort morphology and topography • Image • Correct pathways (2013) As these are the basis of this investigation it was important to understand what defines them, both in the case of Driver’s non-GIS based analysis and through the application of the GIS-based methodology. This understanding could only be achieved by going back to the sites which Driver initially analysed. It was not realistic to investigate the whole of Driver’s study area, consequently a subset of sites was chosen. This chapter also offers the first test of the methodology by testing the effectiveness of the GIS-based methodology compared to the results of non-GIS based analysis (i.e. Driver’s observations).

Situated 8.7km to the south-west is Castell Tregaron which is on a flat topped rocky promontory. This site is surrounded by two banks on its north-eastern side (Driver 2013, 104; Driver 2005a, 277) (Figure 11). These obstruct the eastern entrance, consequently it can only be approached by travelling under the southern end of the outer rampart (RCAHMW 2004). As a result of its obstructive nature Driver argues that the outer rampart is of a later phase to both the entrance and the inner rampart (2005a, 277). Situated 5km to the west is Trecoll. This is an inland promontory fort that is defended by streams and river valleys on three sides. The site’s northeastern side is heavily artificially defended by two large inner banks and an outer rampart (Driver 2005a; Driver 2013) (Figure 12). The south-eastern side of the fort is defined by a pair of low scarp banks which reinforce and emphasise the slope of the topography. There are also traces of stone revetment on the eastern inner most rampart (RCAHMW n.d.b). The site’s entrance is formed by the termination of all three of the eastern/northeastern ramparts in a north-south line to create a gap in the north/ north-west (Ibid. n.d.b).

Data The subset under investigation comprises eight of the sites that exhibit the Pen Dinas and Cors Caron façade schemes (Driver 2013, 145 +147, Figure 9): • The Pen Dinas façade scheme comprises sites which are set along a precipitous edge that forms their longest side (Ibid. 2013, 130). On the long opposing side there are a series of widely spaced terraces, which do not promote direct access (Ibid.). Access is gained through an entrance which is situated on the narrow end of a ridge (Ibid.). These entrances employ ‘exotic’ technology through the use of stone lined passages, bastions and bridges (Ibid.). • The Cors Caron sites are sited around prominent outcrops and ridges, but have very few artificial defences (Ibid., 131). The artificial defences that do exist are situated on the narrow end of the promontory/ ridge and block direct access to the entrance (Ibid.). These defences are steep and close set (Ibid.).

To the north, Gaer Fawr is situated on the summit of a steep east-west ridge that overlooks the Trawsgoed Basin. The site measures 265m north-south by 120m east-west (Driver 2012) and is enclosed by a bivallate rampart in the north-east (Ibid.) (Figure 13). The remaining sides are enclosed by a singular rampart that consists of shale, rubble, and clay, and has traces of a stone revetment (Ibid.). The bank on the site’s western side is less prominent and scarp-like (Ibid.). Gaer Fawr has two entrances, the main one is situated in the east (Driver 2013, 96). This is 23

A GIS-based Analysis of Hillfort Location and Morphology

Figure 9 Distribution of hillforts within the Ceredigion test area (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence), Dyfed Archaeological Trust 2014)

24

Ceredigion

Figure 10 1m resolution LiDAR hillshade models of Pen y Bannau and its immediate environs. Map A focuses on the site, Map B depicts its topographical environs (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

25

A GIS-based Analysis of Hillfort Location and Morphology

Figure 11 1m resolution LiDAR hillshade models of Castell Tregaron and its immediate environs. Map A focuses on the site, Map B depicts its topographical environs (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

26

Ceredigion

Figure 12 1m resolution LiDAR hillshade models of Trecoll and its immediate environs. Map A focuses on the site, Map B depicts its topographical environs (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

27

A GIS-based Analysis of Hillfort Location and Morphology

Figure 13 1m resolution LiDAR hillshade models of Gaer Fawr and its immediate environs. Map A focuses on the site, Map B depicts its topographical environs (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

28

Ceredigion explores the nature and variability of this interaction on a site by site basis.

in-turned and surrounded by a bastion in the east and west (Ibid.). Its other entrance is in the west and is smaller and less elaborate.

Three of the sites, Castell Tregaron, Pen y Bannau and Trecoll exhibit the Cors Caron façade scheme. At Castell Tregaron, the site is defined by the rocky outcrop on which it is situated. The topographical definition of the site is enhanced on its southern and south-western boundaries where the enclosing banks hug the line of the contours. The north-eastern enclosing banks are most prominent and numerous. These face a ‘blind’ hillside, where there is limited visibility beyond it, the remainder of the site faces open downland (Driver 2013, 110). Driver’s observation of the site facing a blind hillside is supported by the results of viewshed analysis as the land to the north east of the site, and within the 1km radius, is not visible (Figure 18). The site has greater visibility of the land within the 3km radius particularly to the west, north-west and aspects of the south-east. Visibility becomes restricted in the 6km radius to the north-west, south-west and scatterings to the south-east and west. Visibility is increasingly scattered to the south-west and north-west in the 9km radius.

To the east and situated on the precipitous northern edge of the Ystwyth Gorge near Pontrhydgroes (Driver 2013, 102) is Castell Grogwynion (Figure 14). This site measures 170m east-west by 100m north-south (Driver 2014). It is arranged into a series of compartments and has two enclosing banks on its northern side, and a singular bank in the east and west. There are three entrances, the main one is situated in the north-eastern corner and is defined by a freestanding walled bastion which is physically combined with an annexe to form an entrance arrangement (Ibid.). The two remaining entrances are less elaborate and are situated in the north and north-west. Tan y Ffordd is situated 7.8km to the north on an overlooked ridge between the Rheidol and Melindwr valleys (Figure 15). The site is largely defined by the topography in the south whilst in the north there are two terraces which follow the line of the enclosing bank. The site’s main entrance is situated in the west (Driver 2005a, 272) where the rampart turns in before it meets the slopes in the south (Driver n.d.) and where gateways are offset from the inner (Ibid.).

Also exhibiting the Cors Coran façade scheme is Pen y Bannau. This encloses an entire outcrop which reduced the amount of effort required to construct an artificial enclosure (Driver 2013, 91; Driver 2005a, 269). The slope of the land is gentlest on the site’s northern side, here the site’s enclosing works are most numerous and prominent. The site’s northern entrance is also placed here which, as Driver highlights, increased the prominence of the entrance due to its ridgeway position (2013).

Located to the west is Pen Dinas. This is situated on a coastal hill and encloses an area of approximately 4ha (Driver 2000). It encloses both the northern and the southern summits of the hill (Figure 16). The northern summit is enclosed by a bank and ditch (Ibid.). This bank is believed to have been timber revetted and had an entrance from the isthmus (Forde, Griffiths et al. 1963). The southern summit was enclosed with a substantial bank and an outer ditch (Driver 2000) with two elaborate gateways, one in the south-east, and the other in the north-east. At a later stage, banks were constructed on the isthmus as too was a stone walled gate which consisted of a two storey building that was crossed by a wooden bridge (Ibid.).

Whilst the topographical position of this site reduced the effort and material required to adequately enclose it, it also promoted a high degree of visual accessibility to the landscape immediately surrounding it. For example, to the 1km radius there is a high degree of all round visibility (Figure 19). Whilst to the 3km radius there is a strong degree of visual accessibility to the west and south-west whereas there is scattered visibility to the remainder of the radius. There is also a strong degree of visibility to the west, south and south-east. This pattern continues towards the 6km radius, but visual magnitude increases to the southeast and east with values in the higher percentile range. To the 9km radius there is a greater spread of visibility but it is become increasingly scattered.

Pen Dinas Elerch is situated to the north-east on an outcrop on the eastern edge of Banc Mynydd Gorddu. It encloses an area that measures 200m north to south by 70m east to west (RCAHMW 2012). On its eastern side the site is defined by two ramparts that enclose a terrace, the western side is defined by a low footing wall which is believed to have originally held a timber revetment (Ibid.) (Figure 17). The main entrance to the site is situated on its southern tip (Driver 2005a, 273; Driver 2013, 118). This opens out onto a ‘causeway’ that gives access to the site from boggy ground (RCAHMW 2012). Another entrance is in the north-western corner of the site, this has a bastion on its southern side (Ibid. ).

Trecoll is situated close to the edge of a river valley escarpment and its morphology largely ignores the topography. For example, the northern enclosing works cut across the land. Whilst the south-eastern enclosing bank very loosely follows the topographical contours. Trecoll’s position does however promote a varied degree of visual accessibility to the surrounding landscape. The site has a high degree of visibility to the 1km radius (Figure 20). This visibility is primarily focused to the south and south-west, and a limited area in the east. Visibility decreases between the 1km and 3km radii, here it is restricted to the west and south-east. Between the 3km and 6km radii visibility

Physical relationship of the hillfort morphology and location with the landscape topography The Pen Dinas and Cors Caron façade schemes interacted with the underlying topography differently. This section 29

A GIS-based Analysis of Hillfort Location and Morphology

Figure 14 1m resolution LiDAR hillshade models of Castell Grogwynion and its immediate environs. Map A focuses on the site, Map B depicts its topographical environs (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

30

Ceredigion

Figure 15 1m resolution LiDAR hillshade models of Tan y Ffordd and its immediate environs. Map A focuses on the site, Map B depicts its topographical environs (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

31

A GIS-based Analysis of Hillfort Location and Morphology

Figure 16 50cm resolution LiDAR hillshade models of Pen Dinas and its immediate environs. Map A focuses on the site, Map B depicts its topographical environs (Map B Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence), Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

Gaer Fawr, another example of the Pen Dinas façade scheme, has morphology that largely concurs with the topography. The only aspect of the site which cuts across the ridge is the eastern defences. The northern corner deflects to take in the summit of the hilltop. Gaer Fawr’s location promotes all round visibility of the surrounding landscape at varying distances. For example, a large proportion of the 1km radius is visible apart from the north-eastern corner (Figure 22). The visibility towards the west, north-west, south and south-west increases to the 3km radius, the south-eastern corner of this area is also visible but to a lesser extent. To the 6km radius, the visible spread increases as scattered areas of the northern half of the radius along with the west, east and south are now visible. There is still a high degree of visibility to the west and north-west in the 9km radius. Discreet areas of the north, north-east and south are also visible.

becomes scattered. The scattered nature of the visibility of the landscape increases between the 6km and 9km radii, this is confined to the south-west and south-east. Pen Dinas exhibits the Pen Dinas façade scheme. This site is situated alongside the coastline and is largely defined by the topography. This location makes this site one of the ‘strongest and most readily defensible hill[s] within the region’ which is one of several factors that Driver (2013, 112) argues make this site a ‘prime location’. Its position, on the western end of a long ridge, meant that the site is very conspicuous and its prominence is accentuated by the earthworks which encircle it (Driver 2013). Whilst Pen Dinas’ location enhances its physical prominence, it restricts the site’s visual accessibility of the surrounding landscape. There is poor visibility to the 1km radius (Figure 21). However, visibility increases beyond this as the majority of the area within the 3km radius is also visible, whereas to the 6km radius the majority of the areas to the west, north-west and south-west are visible. To the 9km radius a greater area of the west is visible, however visibility becomes increasingly scattered to the east.

Another example of the Pen Dinas façade scheme is Tan y Ffordd which is situated on a ridge. The southern aspect of this site is largely defined by the slopes of the ridge thus reducing the amount of physical effort needed to enclose the site (Driver 2013, 91). The site encloses an outcrop in 32

Ceredigion

Figure 17 1m resolution LiDAR hillshade models of Pen Dinas Elerch and its immediate environs. Map A focuses on the site, Map B depicts its topographical environs (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

33

A GIS-based Analysis of Hillfort Location and Morphology

Figure 18 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Castell Tregaron as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Dyfed Archaeological Trust 2014; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Viewshed analysis demonstrates that visibility to the 1km radius from Tan y Ffordd is restricted to the northern, western, and eastern areas (Figure 23). As distance from the site increases, the landscape to the south becomes visible with only the south-eastern and south-western

the north-east which forms what Driver argues to be the most impressive aspect of the site (n.d.). More specifically, he suggests that from the outside the outcrop appears to be a part of the site’s artificial defences, when in fact the outcrop itself is cut in two by a rock cut ditch (Ibid.) 34

Ceredigion

Figure 19 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Pen y Bannau as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Dyfed Archaeological Trust 2014; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Pen Dinas Elerch, also of the Pen Dinas façade scheme, encompasses an entire outcrop, which physically defines the site. The enclosure of an entire topographical feature reduced the amount of effort, and the cost involved in the construction of this artificial enclosure (Driver 2005a, 269; Driver 2013, 91). Whilst the hillfort enhances the

areas of the 3km buffer area being visible. Visibility to the 6km radius is also very restricted as only a small scattering of area in the south-west and south-east beyond the 3km radius are visible. To the 9km, radius visibility is even more restricted and there is only scattered visibility to the south-east. 35

A GIS-based Analysis of Hillfort Location and Morphology

Figure 20 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Trecoll as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Dyfed Archaeological Trust 2014; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

natural form of the landscape, and the topography aided its construction, the topographical position also influenced the visibility of the landscape from the site. To the 1km radius there is very scattered visibility (Figure 24) with the majority of the visible areas being in the west and east. Beyond 1km, and towards the 3km radius, the site’s visibility to the east and north-east increases. Between

the 3km and 6km radii there is increased visibility to the north-east and north-west, whilst between the 6km and 9km radius visibility is restricted solely to the north-west. The final example of the Pen Dinas façade scheme is Castell Grogwynion. This is positioned on the precipitous 36

Ceredigion

Figure 21 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Pen Dinas as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Dyfed Archaeological Trust 2014; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

edge of the Ystwyth Gorge which delineates the site’s southern and western edges. To the north, the enclosing earthworks cut across the lie of the land and are more numerous and prominent. They create a prominent spatial division between the hillfort and the adjacent land. This forms the site’s northern terraced façade, however, the

surface topography and the sub-surface geology inevitably made the construction of this very difficult (Driver 2013, 102). This topographical position also enables the majority of the 1km buffer area to be seen, however visibility decreases at 37

A GIS-based Analysis of Hillfort Location and Morphology

Figure 22 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Gaer Fawr as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Dyfed Archaeological Trust 2014; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Image

the 3km radius, as it is mainly limited to the southern half of this radial area (Figure 25). Visibility becomes more scattered towards the 6km radius because it is confined to the west and south-east. The scattered nature of the visibility from the site increases at the 9km radius as it is confined to the east, west, and north.

Driver highlighted that a number of the hillforts within his study possessed architectural components which potentially served a monumental function. Whilst these components may have been monumental in size, the 38

Ceredigion

Figure 23 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Tan y Ffordd as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Dyfed Archaeological Trust 2014; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

results of viewshed analysis indicates that their size was not proportionate to their potential audience.

argues that it served a monumental function because it is the most visually imposing aspect of the site (2013, 124). The site itself is not readily visible when it is approached from its ‘natural’ approach in the north and north-west however the western outcrop becomes visible within the small valley on the landward side of the site (Ibid.). This gives

For example, at Castell Grogwynion the western outcrop forms one of several ‘compartments’ within the site. This compartment did not enclose habitable space, instead Driver 39

A GIS-based Analysis of Hillfort Location and Morphology

Figure 24 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Pen Dinas Elerch as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Dyfed Archaeological Trust 2014; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Overall, viewshed analysis found that Castell Grogwynion is of varying visibility from the surrounding landscape. Visibility increases as distance from the site increases (Figure 26). For example, from the 1km radius the site’s southern area is the most visible aspect as it has the highest visual magnitude. From the 3km radius the majority of the

the impression that the outcrop forms the entirety of the site (Ibid.). However, the visual prominence of the western outcrop does not directly correlate with its actual visibility. This outcrop, in its entirety, is not the most visible aspect of the site. Instead the south-eastern corner of the outcrop constitutes one of its most visible aspects (Figure 26). 40

Ceredigion

Figure 25 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Castell Grogwynion as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Dyfed Archaeological Trust 2014; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

site’s interior is visible but it is of lower visibility than from the 1km radius. The northern banks are of a lower visibility than the remainder of the site, however, the north-western corner of these earthworks is not visible. This sector of the enclosure is also not visible from the 6km radius. Here, the overall visibility of the site is much

less than from the 1km and 3km radii as the enclosure only has a visual magnitude within the 49th percentile and below. The southern aspect is the site’s most visible area from the 6km radius. The visual magnitude of the entire site increases from the 9km radius and the site’s central southern line is of the highest visibility. 41

A GIS-based Analysis of Hillfort Location and Morphology

Figure 26 Results of viewshed analysis from the radii towards Castell Grogwynion depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (DTM for Viewshed © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

42

Ceredigion

Figure 27 Results of viewshed analysis indicating the visibility of Castell Grogwynion from Pen y Bannau (DTM for Viewshed © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

From the surrounding hillforts within this test area, Castell Grogwynion is of limited visibility and is only visible from Pen y Bannau (Figure 27) which is situated c.6km to the south-east. From Pen y Bannau, the visible areas fall into the relatively high visibility range; however, the site’s

enclosing earthworks are of limited visibility. Although the viewshed analysis suggests that Castell Grogwynion is of high visibility from Pen y Bannau, the distance between the sites inevitably meant that visual clarity was poor, as shown by Ruestes’ analysis of the decay of visual clarity 43

A GIS-based Analysis of Hillfort Location and Morphology mirage is further supported by the selective revetment of the site’s inner bank, which is visible from the land that overlooks the site from the south-east.

with distance. According to Ruestes’ bands of visibility, visibility is possible from distances of 6km however it is sometimes not possible to determine what is being looked at (2008).

GIS-based viewshed analysis found that the site as a whole is of limited visibility from the surrounding landscape (Figure 30). Particularly the north-eastern façade which is not of high visibility, neither is it exceptionally more visible than the remainder of the site. The site as a whole is most visible from the 1km radius as the majority of the enclosure can be seen. From the 3km radius, the visibility of the site decreases with the western side not being visible and the remainder having a visual magnitude within the lower quartile range. The visibility of the site continues to decrease from the 6km radius as the visible areas are confined to a smaller area of the eastern aspect with values within the lower quartile range. There is a slight increase in the site’s visibility from the 9km radius as the visible area increases. The visual magnitude of the northern aspect also increases.

Another site with morphological characteristics of a similar monumental scale is Castell Tregaron. Driver noted the ‘impressive’ and ‘uncompromising’ statement that is made by the site’s north-eastern façade (2013, 87). Although the northeastern bank increases the physical and symbolic prominence of the site, GIS-based viewshed analysis failed to confirm that this aspect of the site is its most visible part (Figure 28). For example, from the 1km radius the majority of the site is visible, and it is generally of moderate visibility, but the north-eastern corner of the site is one of its least visible aspects. From the 3km radius the visual magnitude of the site is highly variable with the highest being the north-western aspect of the interior. This area, along with the centre of the interior, and aspects of the eastern banks also have the highest visual magnitude from the 6km radius. The overall visual magnitude of the site increases from the 9km radius. From this distance the central area and a very small sector of the outer bank remain to be the site’s most visible aspect. The site is not visible from any of the hillforts within the test area.

Driver argues that the siting of Pen Dinas Elerch on a conspicuous knoll meant that it was a visually and physically commanding site (2013). However, it is not visible from the surrounding lowland valleys or from the routes to the west and the coastal plain to the north (Ibid. 2013, 110). The site is, however, visible as one descends from the mountains in the east and it is from here that it appears as one side of an upland plateau (Ibid.). The results of Driver’s field observations are generally supported by those of the GIS-based viewshed analysis. For example, although the majority of the site is visible from the 1km radius the degree to which these areas are visible is highly variable (Figure 31). In general, the most visible aspects of the site are in the north and north-east, and these have a visual magnitude within the upper percentile ranges. From the 3km radius the entirety of the hillfort is also visible but the visual magnitude of the areas within are much lower than from the 1km radius. The most visible area of the site from the 3km radius is its northern aspect. From the 6km radius the overall visible area of the site decreases, however the visual magnitude increases to within the 25th and 49th percentile. The visual magnitude of the visible areas continues to increase from the 9km radius with values between the 50th and 74th percentile. The slopes of the western side of Pen Dinas Elerch are stone revetted and Driver argues that this was directed towards Bwlch-y-ddwyallt (2013), which is a pass that runs west-east towards the site. Driver further relates the orientation of this westerly façade to the easterly façade of Caer Llety Llwyd, which is to the west of Pen Dinas at the end of the Nant y Grogfwyd (a pass) and joins Bwlch-y-Ddwyallt (Ibid.). The orientation of these façades led Driver to suggest that they were contemporary (Ibid.). However, GIS-based viewshed analysis found that Pen Dinas Elerch is not visible from Caer Llety Llwyd which puts doubt on the degree to which these sites could form a visual and immediate influence upon each other. They did not pose as a constant visual reminder to each other.

At Tan y Ffordd, the site’s most physically prominent and elaborate aspect is within the eastern corner. This façade faces away from the entrance in the west, and formed a blind façade that was designed to be seen by those who were descending from the uplands in the north-east (2013 114). Although the eastern façade is the site’s most physically elaborate component, viewshed analysis demonstrates that it is not its most visible (Figure 29). For example, from the 1km radius the entire site is visible; however, the most visually accessible area is the site’s southern aspect and the least visible is the outer northern bank. From the 3km radius the majority of the site is visible and the visual magnitude of the site is consistent with a percentile value of between 50 and 74. The site is not visible from the 6km and 9km radii. Trecoll is overlooked by the uplands which are situated to the south-east of the site. The site’s prominent façade is positioned towards its landward approach in the north (Driver 2013, 88). The inner rampart formed the ‘principal side wall’ as seen from the higher ground in the south-east (Ibid. 2013). This wall is revetted and forms a monumental face to an aspect of the site which can be overlooked from higher ground. The northern terminal of the inner bank also had small blocks of quartz incorporated into its make-up. Instead of investing a large amount of material and effort into creating a truly multivallate site, the morphology is designed to mimic a multivallation sequence by constructing two large free-standing banks that are separated by ditches (Driver 2013, 103). In terms of portraying a particular image within a particular direction, the site’s north-eastern façade creates a much stronger image than the morphology otherwise did. This 44

Ceredigion

Figure 28 Results of viewshed analysis from the radii towards Castell Tregaron depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (DTM for Viewshed © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

45

A GIS-based Analysis of Hillfort Location and Morphology

Figure 29 Results of viewshed analysis from the radii towards Tan y Ffordd depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (DTM for Viewshed © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

46

Ceredigion

Figure 30 Results of viewshed analysis from the radii towards Trecoll depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (DTM for Viewshed © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

47

A GIS-based Analysis of Hillfort Location and Morphology

Figure 31 Results of viewshed analysis from the radii towards Pen Dinas Elerch depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (DTM for Viewshed © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

48

Ceredigion the visual magnitude varies. For example, from the 1km radius the site is of scattered visibility (Figure 37). The western side continues to be the site’s most visible aspect from the remainder of these radii, it also remains to have a visual magnitude in the upper quartile range.

The most physically prominent and elaborate aspect of Pen y Bannau is the northeastern enclosing earthworks. These form the site’s façade and exhibit a falsified image of ‘complete’ multivallation (Driver 2013, 103). This façade, as with that at Castell Tregaron, faces away from the lowlands and towards the mountains in the north-east and according to Driver ‘visually commands the mountain fringe’ although its orientation was highly limited (2013, 115). The limited visibility of the north-eastern façade is supported by the results of viewshed analysis. This façade has the lowest visual magnitude out of all of the site’s architectural components from its surrounding radii. The site as a whole becomes more visible as distance from it increases. From the 1km radius there are varying degrees of visual accessibility to the site with the most visible area being its western side (Figure 32). The northern outer bank is not visible from this distance and the least visible aspect is the eastern side. Beyond this distance and from the 3km, 6km and 9km radii the site’s visibility increases as the majority of the site is of high visibility, although from the 6km and 9km radii the northeastern corner of the site is not visible. Castell Grogwynion is the only hillfort within this test area from which Pen y Bannau can be seen (Figure 33). Only the western side of the site, along with the western side of the northern banks are visible. These areas are of high visibility.

Correct Pathways Morphologically induced correct pathways Entrances by their very principle formed a morphologically defined correct pathway as they were prescribed entry and exit points into sites. However, the key focus within this section are those entrances and morphological components that physically formalised movement and went beyond a simple break in the enclosing circuit. These morphologically induced correct pathways are subsequently compared to the GIS-based correct pathways. At Castell Grogwynion, Driver argued that the site’s main entrance was in the north-east (2013, 74-75). According to him, one ought to approach the site along its northern side and pass southwards along the western side of the bastion. Subsequently one would make a U-turn at the annexe to enter the site’s interior through the entrance (Ibid.). However, none of the GIS based correct pathways correlated with this (Figure 38).

At Gaer Fawr, the site as a whole is highly visible from the surrounding landscape. There is not one distinctly more visible aspect of the site’s morphology than others and this variation is demonstrated both with distance and area (Figure 34). From the 1km radius, the banks which enclose the site are its most visible component and are of higher visibility than the interior. From the 3km radius the overall visual magnitude and the actual visible area increases as the majority of the site is of high visibility. This continues from the 6km radius, however from the 9km radius only the northern and southwestern area of the site is of high visibility. The remainder of the site is of low visibility, or not visible.

Driver also postulated a morphologically induced correct pathway at Tan y Ffordd. Here he noted that the elaboration of the site’s entrance was directed towards traffic approaching the site whilst ascending a long slope (Driver n.d.). This traffic would have proceeded to enter the outer gateway past a ‘command post’ and then undertaken a sharp reverse ‘s’ by travelling uphill to the right, and then sharply left through the inner gate (Ibid.). However, just as with Castell Grogwynion there was no correlation between the GIS-based least cost pathways and this morphologically defined path (Figure 39).

Situated 10.5 km to the south-east of Gaer Fawr is Pen Dinas. Driver argues that Gaer Fawr is not intervisible with Pen Dinas (Driver 2005a, 271; Driver 2013, 96). However, the results of GIS-based viewshed analysis from both the Profile DTM and LiDAR based DTM found that these two sites are intervisible (Figure 35). Even though the results of viewshed analysis indicate that these sites are intervisible, only seven of the points within the Pen Dinas can see Gaer Fawr, and Gaer Fawr can only see the north fort at Pen Dinas. (Figure 36). It is highly likely that the variation in the viewshed results to those of Driver’s observations is due to the placement of the points in comparison to where Driver was stood. A point placed in the ‘wrong’ position or Driver having stood in the ‘wrong’ position inevitably meant that observations would differ. This emphasises the importance of using both viewshed analysis and field visits to come to balanced conclusions.

At Pen Dinas Elerch a causeway was constructed to enable access into the site from the boggy lowlands. The slopebased least cost pathways have a close association with this area, a total of 251 of the slope-based pathways use it. The visible pathways also have a strong correlation with this southern area. Driver also argued that the lowland pass of Bwlch y Ddwyallt would have been an important routeway in the past (2013). However, slope-based cost surface analysis indicated that only 2 of the 257 least cost pathways use this pass (Figure 40). Castell Tregaron is situated to the east of, and overlooks, the overland route and valley junction of the rivers Berwyn and Groes with the Teifi. GIS-based cost surface analysis does not coincide with these influential routeways. Driver also argued that another important route into the site was from the north-east (Ibid., 116), although again, a limited number of least cost pathways approach the site from this direction (Figure 41).

At Pen Dinas, the most consistently visible aspect of the site is its western side, however as distance varies

49

A GIS-based Analysis of Hillfort Location and Morphology

Figure 32 Results of viewshed analysis from the radii towards Pen y Bannau depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (DTM for Viewshed © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

50

Ceredigion

Figure 33 Results of viewshed analysis indicating the visibility of Pen y Bannau from Castell Grogwynion (DTM for Viewshed © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reservedinformation © Natural Resources Wales and database right. All rights reserved)

This implies that they were constructed within one of the most accessible aspects of these sites.

Slope-based correct pathways Cost surface analysis indicated that six out of the eight hillforts within this test area had at least one entrance which correlated with the slope-based least cost pathways.

For example, at Castell Grogwynion the northern entrance correlates with 250 out of the 257 slope-based pathways 51

A GIS-based Analysis of Hillfort Location and Morphology

Figure 34 Results of viewshed analysis from the radii towards Gaer Fawr depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (DTM for Viewshed © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

52

Ceredigion

Figure 35 Visibility of Gaer Fawr from Pen Dinas. Map A depicts the viewshed results when using the Ordnance Survey PROFILE DTM, Map B depicts the results when using the 1m resolution LiDAR (DTM for Viewshed Analysis in Map A © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

At Pen y Bannau, it was argued that the north-eastern entrance is easier to access than the western entrance which faces steep slopes (RCAHMW n.d.a). This is supported by the results of cost surface analysis which identified that the north-western entrance was placed in one of the site’s most accessible aspects. Here 128 out of 260 slope-based pathways converged (Figure 43).

(Figure 38). Another site with such a strong association between the slope-based least cost pathways and an entrance is Pen Dinas Elerch. Here 251 of the slope-based least cost pathways intersect with the southern entrance (Figure 40). Similarly, at Castell Tregaron, cost surface analysis indicates that the south-eastern entrance was placed in the site’s most accessible area as 246 of the 258 slope-based pathways converge here (Figure 41). This reinforces Driver’s point that the addition of the outer bank at Castell Tregaron would have inhibited entry into the site as people had to approach via a less accessible route (2005a, 277).

Moving away from the correlation of the slope-based least cost pathways with site entrances, there are several instances whereby the disproportionate vallation sequences correlated with the most accessible areas. For example, at Trecoll the site’s north-eastern façade and its entrance correlate with 113 of the 257 slope-based least cost pathways (Figure 44). This evidence supports Driver’s statement that the façade faces the site’s landward approach. Similarly, the most accessible aspect of Tan y Ffordd, according to the results of the slopebased cost surface analysis, is the north-western sector. 203 out of the 257 slope-based pathways intersect with this area which is just to the north of the western entrance (Figure 39). At Pen Dinas the slope-based cost surface analysis indicated that the eastern side of the isthmus was the most accessible route into the site as 250 out of 258 slope-based least cost pathways (Figure 45) correspond with this area.

At several sites the correlation of the slope-based least cost pathways is not as strong but it continues to shed light on further potential access related motivations to the morphological orientations of hillforts. For example, at Gaer Fawr the easiest way to enter or exit the site is via the western entrance. Here 130 of the 256 slope-based pathways converged (Figure 42). Paths are also incised into this hilltop from the Trawsgoed Lowlands in the east, however only one slope-based, six blind and one visible least cost pathway corresponds with an easterly approach to the site. 53

A GIS-based Analysis of Hillfort Location and Morphology

Figure 36 Visibility of Pen Dinas from Gaer Fawr. Map A depicts the viewshed results when using the Ordnance Survey PROFILE DTM, Map B depicts the results when using the 1m resolution LiDAR (DTM for Viewshed Analysis in Map A © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

that their morphology was oriented towards an area where the site had most visibility to the surrounding landscape. At Trecoll the visible pathways strongly coincide with the site’s northern aspect and its nearby entrance as 25 out of 41 interact with this area (Figure 46).

Visible correct pathways Above, the potential for the results of slope-based analysis to identify instances whereby the physical accessibility of particular aspects of sites influenced their morphological orientation was discussed. Here the correlation of the visible pathways with the morphological orientation of sites is explored to examine the evidence for hillforts potentially orientating their morphology based upon the visibility of the surrounding landscape.

Two further hillforts have entrances that coincide with the majority of the visible pathways. At Gaer Fawr the visible pathways have a high correlation with the western entrance as 9 of them use this area (Figure 48). Similarly, the visible pathways have a strong correlation with the southern entrance at Pen Dinas Elerch as 13 out of the 42 coincide with this area (Figure 49).

The most prominently enclosed aspects of both Trecoll and Castell Grogwynion correspond with the majority of their visible pathways (Figure 46 and 47). This indicates 54

Ceredigion

Figure 37 Results of viewshed analysis from the radii towards Pen Dinas depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (DTM for Viewshed © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

55

A GIS-based Analysis of Hillfort Location and Morphology

Figure 38 Results of slope-based cost surface analysis at Castell Grogwynion (DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

aspect of the site correlates with 12 of the 42 visible pathways (Figure 50). At Pen y Bannau there is a strong correlation between the visible pathways and the southwestern and western aspect of the site (Figure 51). At Pen Dinas the majority of the visible pathways correspond with

Alternatively, at Tan y Ffordd, Pen y Bannau and Pen Dinas although there is a strong correlation of the visible pathways with particular aspects of the site, they do not correspond with the most enclosed aspect or any of the entrances. For example, at Tan y Ffordd the south-eastern 56

Ceredigion

Figure 39 Results of slope-based cost surface analysis at Tan y Ffordd (DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

Blind correct pathways

the eastern side of the isthmus (Figure 52). The majority of these travel to/from the site from both the uplands and lowlands to the east. The span of these pathways strongly indicates that this site has a high degree of visibility to the area immediately to the east.

Blind pathways outlined areas from which the hillforts were least visible. Assessing the correlation of the morphological components with these pathways therefore identifies areas 57

A GIS-based Analysis of Hillfort Location and Morphology

Figure 40 Results of slope-based cost surface analysis at Pen Dinas Elerch (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

58

Ceredigion

Figure 41 Results of slope-based cost surface analysis at Castell Tregaron (DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

corresponds with 20 out of the 42 least cost pathways (Figure 53). Similarly, at Pen y Bannau the northeastern entrance coincides with 18 out of the 42 blind pathways demonstrating that the entrance was of limited visibility on approach (Figure 54). This area also corresponds with the façade which, coupled with the northern slopes of the crag,

whereby the components had a limited visual audience. This study has identified a strong correlation between these pathways and some of the strongest morphological components of these sites. At both Gaer Fawr and Pen y Bannau the blind pathways had a strong correlation with one of their entrances. At Gaer Fawr the eastern entrance 59

A GIS-based Analysis of Hillfort Location and Morphology

Figure 42 Results of slope-based cost surface analysis at Gaer Fawr (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

60

Ceredigion

Figure 43 Results of slope-based cost surface analysis at Pen y Bannau (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

61

A GIS-based Analysis of Hillfort Location and Morphology

Figure 44 Results of slope-based cost surface analysis at Trecoll (© Crown Copyright and Database Right (2018). Ordnance Survey (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

62

Ceredigion

Figure 45 Results of slope-based cost surface analysis at Pen Dinas (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

63

A GIS-based Analysis of Hillfort Location and Morphology

Figure 46 Distribution of visible pathways at Trecoll (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

64

Ceredigion

Figure 47 Distribution of visible pathways at Castell Grogwynion (DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

At Castell Tregaron cost surface analysis supported the results of viewshed analysis whereby the site’s northern aspect is of poor visibility from the surrounding landscape as the majority of the blind pathways use this aspect of the site (Figure 55). Similarly, at Trecoll the north-eastern façade strongly correlates with 21 of the 42 blind pathways (Figure

inevitably obstructed visibility into the site from the areas immediately outside of it. The results of GIS-based analysis are also supported by Driver’s observations. He highlighted that there was limited visibility of the site’s façade, particularly from a distance as there are several instances where it is obscured from higher ground (2013, 115). 65

A GIS-based Analysis of Hillfort Location and Morphology

Figure 48 Distribution of visible pathways at Gaer Fawr (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

66

Ceredigion

Figure 49 Distribution of visible pathways at Pen Dinas Elerch (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

67

A GIS-based Analysis of Hillfort Location and Morphology

Figure 50 Distribution of visible pathways at Tan y Ffordd (DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

56). At Pen Dinas Elerch the site’s western façade has a strong correlation with the blind pathways as 15 of the 42 least cost pathways interact here (Figure 57). These also travel primarily from the north-west which supports Driver’s observation that the site was not visible from the lowland valleys, particularly to the west and the coastal plain to the north (2013, 110). The

high visibility of the site from the east (Ibid. 2013, 110) was indirectly supported by the cost surface analysis because no blind pathways travelled to or from this direction. At Tan y Ffordd the least visible area when approaching from neighbouring hillforts is in the south as 21 out of 68

Ceredigion

Figure 51 Distribution of visible pathways at Pen y Bannau (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

69

A GIS-based Analysis of Hillfort Location and Morphology

Figure 52 Distribution of visible pathways at Pen Dinas (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

70

Ceredigion

Figure 53 Distribution of blind pathways at Gaer Fawr (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

71

A GIS-based Analysis of Hillfort Location and Morphology

Figure 54 Distribution of blind pathways at Pen y Bannau (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

72

Ceredigion

Figure 55 Distribution of blind pathways at Castell Tregaron (DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

42 blind pathways meet there (Figure 58). The restricted visibility when approaching from this area is caused by the site’s ridge end position. However, the overall visibility when approaching from the west is also limited as seven of the blind paths enter the site very close to the western entrance.

The wider archaeological context Whilst the GIS based analysis identified multiple potential correct pathways, from a site perspective it is important to relate these to the known contemporary landscape. However, this correlation only occurred with some of the 73

A GIS-based Analysis of Hillfort Location and Morphology

Figure 56 Distribution of blind pathways at Trecoll (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

74

Ceredigion

Figure 57 Distribution of blind pathways at Pen Dinas Elerch (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

75

A GIS-based Analysis of Hillfort Location and Morphology

Figure 58 Distribution of blind pathways at Tan y Ffordd (DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

(PRN-12034; PRN-12027) and two Bronze Age standing stones (PRN-12033; PRN-12026) (Figure 59). The dense concentration of activity within this area coupled with the high correlation of several pathways implies that this was a significant area within later prehistory, perhaps a navigational point through the landscape.

pathways at Pen Dinas Elerch. Here, the overall results of cost surface analysis demonstrate that a large number of pathways travel through the land to the south of Pen Dinas Elerch, an area that is also relatively densely populated with a number of archaeological features at Banc MynyddGorddu. These include two potential Bronze Age Barrows 76

Ceredigion

Figure 59 Results of cost surface analysis indicating the correlation between the least cost pathways and known areas of activity at Pen Dinas Elerch. Map A depicts the results of slope-based cost surface analysis, Map B depicts the distribution of the visible pathways and Map C depicts the distribution of the blind pathways (Dyfed Archaeological Trust 2014; DTM for cost surface analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence)

77

A GIS-based Analysis of Hillfort Location and Morphology Conclusion

that this can be disproven by those who were visiting the site. Here, this risk was minimized through the placement of the façades where there was limited visibility to the remainder of the site. This is an important observation which is explored further in Chapter 4 and 5.

The application of a GIS-based methodology to a subsection of Driver’s study area enhanced his observations due to GIS being able to investigate large quantities of data across large areas, which would not feasibly be possible within the time frame of an individual’s research. Driver’s work was limited to the routeways that he chose to walk which were pre-defined topographical routes such as passes which meant that his study did not allow for the investigation of routes which were not topographically defined. Topography is likely to have defined movement in the past; however, people would also have moved around the landscape beyond these routeways thus forming a part of their visual understanding of these sites. Consequently, by investigating visibility on a radial and site basis this GISbased study reduced the impact that choices of observation points had upon the overall visual understanding of sites.

The comparison of Driver’s field observations with the results of GIS based analysis highlighted the complexity in terminology, particularly that describing the visibility and impression of hillforts and their enclosing works. Multiple examples were identified where Driver termed an architectural component as a site’s most visible aspect, when in fact viewshed analysis showed that they were not. Instead these areas were visually prominent and such prominence could have been caused by the contrast between the form of these components with the remainder of the site and the surrounding topography. Driver’s experience of the sites within their landscape context influenced his visual interpretation of them. This was probably caused by the visual prominence of the architectural components and their ability to attract the eye, whereas GIS based analysis provides a more objective and quantifiable insight into a site’s visual qualities. This is explored further in Chapters 4 and 6 where further test area analyses highlight the contradictory nature of fieldwork observations and viewshed analysis.

There were some differences between the results of visibility analysis and Driver’s observations in the investigation of the concept of image. For example, Driver claimed that Gaer Fawr and Pen Dinas were not intervisible, whereas viewshed analysis showed that they are. However, the visible areas are of low visibility which meant that only a limited number of points on the hillfort grid (on either site) can see the area. The failure of the viewshed results to support Driver’s observation could simply be that he was standing in the wrong place at the wrong time. The viewshed results also failed to support Driver’s observations in regard to Trecoll’s façade, which he believed were designed to face its landward approach. The façade as a whole was of limited visibility from the surrounding landscape. It did not have the visual impact that was implied, even though the physical construction of the façade was an attempt to manipulate the site’s image. This was also mirrored at Castell Grogwynion, Castell Tregaron, Pen Dinas Elerch and Tan y Ffordd where the most prominently enclosed aspects of these sites were not necessarily the most visible.

In terms of correct pathways there was limited agreement between the results of cost surface analysis and Driver’s topographically and morphologically defined pathways. For example, none of the least cost pathways follow the complex entrance arrangement at Castell Grogwynion, Castell Tregaron or Tan y Ffordd which were identified by Driver. However, at Gaer Fawr and Pen y Bannau there were multiple instances where several types of least cost pathways intersect with the morphologically defined correct pathways as indicated by the entrance placement. Similarly, there are several instances where the slopebased least cost pathways in particular at Pen Dinas Elerch intersect with the causeway to the south of the site. The concept of correct pathways and the supporting evidence is discussed extensively in Chapter 5.

The poor visual impact of these was also amplified by the results of Cost Surface Analysis. This identified multiple instances where blind pathways intersect with the most elaborate aspect of a number of sites. The strong correlation of the blind pathways with the façades at Castell Tregaron, Trecoll and Pen y Bannau indicate that they faced an area where the site’s interior is of limited visibility upon approach. The existence of these façades manipulated site images. However, the risk of creating a false impression is

The results of the investigations within this chapter have raised issues surrounding terminology, objectivity, and the correlation of blind pathways with prominent façades which are issues that will be investigated further in the subsequent chapters. Chapter 4 continues to explore the physical relationship between the hillfort morphology and topography, image and the evidence for correct pathways, by applying the GIS-based methodology to hillforts in Aberdeenshire, Gower, Warminster and Dartmoor.

78

4 Multi-regional GIS-based Analysis of Hillfort Location and Morphology Introduction to the test areas

Prestonbury Castle is situated 1.5 km to the north-west of Wooston Castle (Figure 63). This sits at 241m OD and overlooks the gorge of the River Teign. It consists of three enclosures which all have entrances on their eastern side. The inner enclosure also has an entrance on its western side (Historic England, 2015b). The inner and middle entrances are simple gaps whilst the outer entrance in the north-east is in-turned (Forde-Johnston 1976, 221). It is possible that the outer enclosure was never finished (Historic England, 2015b). The outer earthwork is a single curving cross bank which adds another 4.8ha to the hillfort enclosure (Ibid.). The inner two enclosures enclose 1 ha (inner) and 0.8ha (middle) (Forde-Johnston 1976, 188).

Chapter 3 detailed the results of applying the methodology to a subset of Driver’s study sites in Ceredigion. This formed the first test of the affectivity of the GIS-based methodology particularly as it was a key comparison between a GISbased methodology and field-based observations. To test the methodology further, this chapter details the results of its application to four test areas which are both topographically variable and possess hillforts of a different morphological character. This section introduces the test areas, their topographical character and the hillforts within them. Dartmoor

Aberdeenshire The most topographically distinctive region in the study is the Dartmoor test area. This is an upland moorland landscape that is dominated by the steep sided valley of the River Teign (Figure 60). This valley drew people to the area and was evidently significant enough for people to significantly invest in it through the construction of sites such as the hillforts under study. The hillforts within this test area are all hillslope enclosures that are located on the edges of the river valley and were separated by a distance of between c.1 km and c.3 km. The sites in question are: Cranbrook Castle, Prestonbury Caste and Wooston Castle.

Another topographically distinctive region is the Aberdeenshire test area, this is dominated by several hill ranges. It is densely occupied by a series of waterways but has a very low concentration of hillforts. Four of the hillforts within this region are under study, three of which are located in a cluster on the western side of the test area, the remaining site is on the eastern side (Figure 64). The cluster comprises Tap o’ Noth, Cairnmore and Wheedlemont whilst the eastern site is Dunnideer. These hillforts enclose areas which range from c.0.3ha to c.21ha. The distance between them ranges from c.3km to 14km.

Cranbrook Castle consists of a stone-built bank with very slight ditches (Newman 2011) on all sides apart from the north. A ditched counterscarp bank is also present on the site’s southern and south-western side (Historic England 2015a). A slight trace of a bank was identified in the north through LiDAR analysis (Figure 61). There are two entrances, a simple gap with a causeway in the east and a causewayed gap with a curved extension to the ditched counterscarp bank in the south-west (Ibid.). Although no absolute dating has been undertaken at Cranbrook Castle, chronological relationships are exhibited. Cranbrook is centrally situated within a complex of reaves. These reaves descend from the hilltop and are believed to be a response to land pressure (Fleming 1996, 65 +70).

Two of the sites have a vitrified inner enclosure, these are Tap o’Noth and Dunnideer. Dunnideer is sited at 265m OD on a conical shaped hill (RCAHMS 1990). The site also has a Medieval tower (Ibid.). The fort consists of multiple enclosing banks (Figure 65). Two of these lines are situated on the summit of the hill, an intermittently trivallate sequence is also situated downslope. At the western end of the summit enclosure there is a potential contemporary cistern (Ibid.). Three programmes of dating have taken place at Dunnideer. In 1989 Gentles undertook archaeomagnetic dating on the north-eastern corner of the inner vitrified wall. In 2010, a further scheme of archaeomagnetic dating was undertaken by Harris and Hounslow on the inner vitrified wall. Additionally, radiocarbon dating was carried out on the inner vitrified wall in 2008 (Cook 2010a).

Located 2.5 km to the east of Cranbrook Castle is Wooston Castle (Figure 62). This consists of three outer cross banks that face uphill and define the slopes of the spur (FordeJohnston 1976, 206). The second short section of bank is positioned either side of an in-turned entrance which adjoins a hollow way that runs between the second and third banks (Quinnell, Griffith et al. 2013). A hollow way also potentially ran between the third and fourth bank. This is formed by the continuation of the eastern sector of the third ramparts in a northerly direction towards the inner enclosure (Ibid.).

The results of Gentles’ archaeomagnetic dating on the inner vitrified enclosure were inconclusive due to their large spread (1989). However, the results from the 2010 archaeomagnetic dating were more conclusive. Six samples, that were believed to be in situ, were taken from the south-eastern portion of the vitrified wall (Harris and Hounslow 2010). This suggested that the ‘date of 79

A GIS-based Analysis of Hillfort Location and Morphology

Figure 60 Distribution of hillforts within the Dartmoor test area (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Dartmoor National Park Historic Environment Record 2013)

last heating was 345 BC (95% confidence interval of 606-257 BC)’ (Ibid.).

interior of the inner fort, which meant that they provide a date of occupation within the inner fort (Ibid.).

The 2008 radiocarbon dating programme retrieved several radiocarbon dates from charcoal taken from burnt wood from within the collapsed rampart (Cook 2010a). These samples had dates of 390-190 BC and 370-160 BC (Ibid.). Cook hypothesised that the latter sample may represent wood which was brought to the site as fuel for vitrification, or it may have been from a collapsed wooden structure (Ibid., 85). However, he was sure that it did not derive from the timber lacing of the rampart (Ibid.). The dates that derive from the charcoal samples both predate the firing of the inner rampart (Ibid., 86). It is also assumed that they derived from structures that were within the

Tap o’ Noth is situated at 564m OD and encloses an area of 21ha (RCAHMS 2007). The site is enclosed by two walls (Figure 66). The inner wall, which encloses the summit of the hill, consists of a stone and an inner timber laced wall; the wall is heavily vitrified and was originally c.6.1m thick (RCAHMS n.d.). There is a medial ditch between the two banks that cut across the south-eastern corner of the inner fort (Dunwell and Strachan 1997, 11; Halliday 2008, 105). This ditch potentially derives from an earlier enclosure on the hilltop (Ibid. 2008, 105). Lower down the hill a second lesser stone bank delineates another fort (RCAHMS n.d.). There is no evidence of an entrance at this site, but this has 80

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 61 Hillshade model of 1m resolution LiDAR DTM for Cranbrook Castle. Map A focuses on the site, Map B depicts its topographical environs (Dartmoor National Park Historic Environment Record 2013; © Environment Agency copyright and/or database right 2015. All rights reserved; Basemapping for Map B © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

81

A GIS-based Analysis of Hillfort Location and Morphology

Figure 62 Hillshade model of 1m resolution LiDAR DTM for Wooston Castle. Map A focuses on the site, Map B depicts its topographical environs ( Dartmoor National Park Historic Environment Record 2013; © Environment Agency copyright and/or database right 2015. All rights reserved; Basemapping for Map B © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

82

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 63 Hillshade model of 1m resolution LiDAR DTM for Prestonbury Castle. Map A focuses on the site, Map B depicts its topographical environs (Dartmoor National Park Historic Environment Record 2013; © Environment Agency copyright and/or database right 2015. All rights reserved; Basemapping for Map B © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

83

A GIS-based Analysis of Hillfort Location and Morphology

Figure 64 Distribution of hillforts within the Aberdeenshire test area (Aberdeenshire Council Archaeology Service 2014; Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

1988). Additionally, archaeomagnetic dating was carried out on material from the north-western aspect of the inner vitrified wall (Gentles 1989). The thermoluminescence dating retrieved a terminus post quem date for the vitrification of 2160BC+_180+_410 (Sanderson, Placido, Tate 1988, 315). However, the archaeomagnetic dating retrieved dates from the 1st Century BC to the 1st Century AD (Gentles 1989). There is evidently no agreed date for Tap o’ Noth. Cook postulates that, based upon the site’s morphology, its outer wall could have been constructed

been a contentious issue. Hibbert placed the entrance in the south-east and argued that it had a series of outworks (1857, 296). Present-day access to this site is gained by travelling over debris in the south-east. There are ten gaps within the outer bank which mainly occur in the north (Halliday 1999; 2008, 105). Tap o’Noth has been subject to two dating programmes. Thermoluminescence dating was undertaken on vitrified material from the inner wall (Sanderson, Placido, Tate 84

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 65 Plan of Dunnideer (Crown Copyright: Historic Environment Scotland)

Figure 66 Plan of Tap o’ Noth (Crown Copyright: Historic Environment Scotland)

85

A GIS-based Analysis of Hillfort Location and Morphology

Figure 67 Plan of Wheedlemont (Crown Copyright: Historic Environment Scotland)

(Figure 69). The topography is predominantly flat coastal plain, which also has a series of isolated hills and ridges such as Llanmadoc Hill, Hardings Down, Rhosilli Down and Cefn Bryn. The sites under investigation are The Bulwark, which is located on the eastern terminus of Llanmadoc Hill, and the Hardings Down North, West and East Enclosure.

in the late Bronze Age whilst the inner wall was likely constructed within the Middle Iron Age (2013). The remaining two hillforts within the Aberdeenshire test area are both situated on conical shaped hills with a simple enclosure. Wheedlemont is an oval shaped site that is situated on the summit of Wheedlemont Hill (Figure 67). The site’s entrance is situated in the north-east and is 3m wide (Ordnance Survey 1967). Although scientific dating has not been carried out at Wheedlemont, Cook has postulated an Early Medieval date for it (2013). This date was extrapolated from sites of a known date and with a similar morphology and spatial location (Ibid.).

The Bulwark is a hillslope fort, that has two enclosures (Figure 70). The inner enclosure is defined by a bank and ditch with steep slopes to the north, whilst the outer is defined by the inner and outer banks (RCAHMW 1976). The site’s entrance is situated on its eastern side and is approached by a 12m wide passageway (Ibid.). The entrance is a simple gap and penetrates from the third set

Cairnmore consists of an outer soil bank and ditch, a middle stone wall and a lesser inner stone wall (Cook 2010b, 63) (Figure 68). In 2010, Cook undertook a series of key-hole excavations at the site and found that it consisted of a pair of enclosures (2012, 8). He also discovered a third outer bank and ditch (Ibid.). Radiocarbon dating has been carried out at Cairnmore. These dates were retrieved from some charcoal that was found underneath the middle rampart, this had a radiocarbon date of 1510+-30 BP (Cook 2013). The other was from some charcoal from within a destruction layer over a rampart and this had a radiocarbon date of 1580 +-30 BP (Ibid). The Gower Unlike the Dartmoor and Aberdeenshire test areas the Gower is a less topographically prominent region. This consists of an inland section of the Gower Peninsula

Figure 68 Plan of Cairnmore (Ordnance Survey data © Crown copyright and database right 2012)

86

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 69 Distribution of hillforts within the Gower test area (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014))

Record 2013) (Figure 71). The site’s entrance is situated in the north-east of the main inner enclosure and forms a break in the enclosure circuit. This is accompanied by a slight out-turning of the rampart and a four-post timber framework that measures 9ft square (Hogg 1973, 59). A cobbled roadway was also situated between the kerbs of the entrance (Ibid., 60).

of banks and ditches (Ibid.). Approximately 68m away from the western outer bank of The Bulwark is a ditch which cuts across the entire hilltop (north-south). The date of this ditch is unknown. Although no absolute dating has taken place at The Bulwark, the physical evidence on Llanmadoc Hill indicates that there are distinct chronological relationships on the hill. For example, The Bulwark is extensively enclosed on its western side where the site faces the remainder of Llanmadoc Hill and the barrows that are located along the hilltop.

Hardings Down North Enclosure is situated on the northern slopes of Harding’s Down and encloses 0.2ha (Glamorgan Gwent Archaeological Trust Historic Environment Record 2013) (Figure 72). Its enclosure comprises a bank, ditch and counterscarp with an entrance in the north-west (Ibid.).

Hardings Down West Enclosure encloses 0.6ha (Glamorgan Gwent Archaeological Trust Historic Environment 87

A GIS-based Analysis of Hillfort Location and Morphology

Figure 70 Hillshade model of 2m resolution LiDAR DTM for The Bulwark. Map A focuses on the site, Map B depicts its topographical environs (Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved ; Basemapping for Map B © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

88

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 71 Hillshade model of 2m resolution LiDAR DTM for Hardings Down West Enclosure. Map A focuses on the site, Map B depicts its topographical environs (Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; Basemapping for Map B © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

89

A GIS-based Analysis of Hillfort Location and Morphology

Figure 72 Hillshade model of 2m resolution LiDAR DTM for Hardings Down North Enclosure. Map A focuses on the site, Map B depicts its topographical environs (Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; Basemapping for Map B © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

90

Multi-regional GIS-based Analysis of Hillfort Location and Morphology prior to the construction of Scratchbury and Battlesbury (Ibid., 78). Excavations on the spur of land to the north of Battlesbury also identified an Iron Age settlement (Battlesbury Bowl settlement) (Ellis et al. 2008). This settlement was largely unenclosed and is likely to have spread across the ridge top and plateau up to Battlesbury Hill (Ibid., 133). However, the construction of the hillfort in the mid to late Iron Age led to the contraction of this settlement (Ibid., 134).

This entrance is approached by a 3m wide hollow way that is set between low stony banks (Ibid.). Hardings Down East Enclosure encloses c.0.9ha (Ibid.) (Figure 73). Superficially this site appears incomplete. However, RCAHMW suggest that the short, isolated stretch of western rampart defines the western perimeter of the site (Forde-Johnston 1976). The western bank is visible on the LiDAR hillshade model. Forde-Johnston believed that the site comprised an oval main enclosure and an annexe (Ibid., 202). The annexe is located to the east of the entrance, which consists of a break in the enclosure circuit. Its southern side has a thickened terminal (Ibid. 1976).

Scratchbury is situated to the south-east of Battlesbury, it is believed to be the predecessor to Battlesbury (McOmish, Field et al. 2002). This univallate enclosure is located on the southern extension of Salisbury Plain at 196m OD. It encloses 17ha and has three entrances which are situated within the east, south-east and west (Historic England 2015e) (Figure 77).

Warminster Another topographically subtle landscape is the Warminster test area which surrounds the town of Warminster in West Wiltshire. This is situated on the outskirts of the high chalk plain extension of Salisbury Plain (Land Use Consultants 2005) and is a rolling chalk downland that has very few distinctive topographical features (Figure 74). Within the centre of the test area there is a very flat plain that borders the edge of the slight uplands of Salisbury Plain. The sites under study are Cley Hill, Battlesbury and Scratchbury, these are separated by greater distances than the Gower sites as the distance between them ranges from c.2km and c.7km.

Other chronological relationships illustrated at this site include the D-Shaped enclosure in the centre. At a later stage, one side of this enclosure was extended by the lengthening of the straight side towards the hillfort defences; this was later abandoned in favour of using the hillfort’s outer defences (Forde-Johnston 1976, 107). The D-shaped enclosure was also truncated by a linear scarp of unknown origin and function, which is overlain by the hillfort defences close to the north-eastern entrance (McOmish et al. 2002, 75). Excavations of this feature revealed that it had a ditch on its eastern side; this feature is argued to be a cross-ridge bank that preceded the construction of the hillfort (Ibid., 75-76). There are also six to seven round barrows within the site’s interior (Colt Hoare 1821, 70, Historic England 2015e).

Cley Hill is a univallate hillfort that occupies a prominent outcrop. More recently the site has been damaged by small scale quarrying and therefore its enclosing works do not appear complete (Figure 75). Colt Hoare did however note that the site was ‘surrounded by a ditch and rampart, bearing the marks of high antiquity’ (1821, 51). This implies that at some point the site was completely enclosed. It has no obvious entrance (Historic England 2015c). There are two barrows in the centre of the site, Grinsell argued that they were Bronze Age in origin (1958); whilst Colt Hoare believed the hillfort was earlier in date (1821).

Summary The topographical character of these test areas is highly variable. Both the Dartmoor and Aberdeenshire test areas are highly prominent regions, the former being dissected by the distinct Teign River valley and the latter populated by prominent hill ranges. The Gower and Warminster test areas are less topographically distinct. The Gower comprises the inland lowlands within the coastal peninsula, whilst the Warminster test area consists of the perimeter of Salisbury Plain. Whilst the topographical character of these regions is variable, the spatial distribution of the hillforts within them is also highly variable. The hillforts are separated by distances that range from over 10km in the Aberdeenshire test area, to less than a kilometre in the Gower test area. The variability in the test cases is also accentuated by their variable character which range from hillslope enclosures to vitrified forts and multivallate hillforts that enclose over 9ha.

Battlesbury is situated to the north-east of Cley Hill, at 196m OD. This multivallate hillfort encloses 9.7ha (Historic England 2015d) and is a contour fort (McOmish et al. 2002, 77). It has two entrances, one in the north-west and the other in the south-east (Figure 76). Three barrows are located within the southern corner of the site. These are located between the inner rampart and the internal lynchet (Historic England 2015d). The two smaller barrows were crossed by the inner rampart (Colt Hoare 1821, 68). Even though these barrows were constructed prior to the construction of the hillfort, the hillfort builders respected them, as they did not use the material from them to strengthen the inner rampart (Ibid., 69). There are also traces of Celtic field systems, which pre-date the hillfort ramparts, within the interior of the site (McOmish et al. 2002, 78).

None of the hillforts have been firmly dated. However, the majority of them exhibit some form of chronological relationship whether that is with the barrows, field systems and enclosures within their interior, the neighbouring hillforts and enclosures, or the selection of absolute dates that have been retrieved. This information, although

Celtic field systems are also recorded in the vicinity of the hillfort (McOmish et al. 2002, 78). These were established 91

A GIS-based Analysis of Hillfort Location and Morphology

Figure 73 Hillshade model of 2m resolution LiDAR DTM for Hardings Down East Enclosure. Map A focuses on the site, Map B depicts its topographical environs (Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; Basemapping for Map B © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

92

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 74 Distribution of hillforts within the Warminster test area (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)

93

A GIS-based Analysis of Hillfort Location and Morphology

Figure 75 Hillshade model of 1m resolution LiDAR DTM for Cley Hill. Map A focuses on the site, Map B depicts its topographical environs (© Environment Agency copyright and/or database right 2015. All rights reserved; Basemapping for Map B © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)

94

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 76 Hillshade model of 1m resolution LiDAR DTM for Battlesbury. Map A focuses on the site, Map B depicts its topographical environs (© Environment Agency copyright and/or database right 2015. All rights reserved; Basemapping for Map B © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)

95

A GIS-based Analysis of Hillfort Location and Morphology

Figure 77 Hillshade model of 1m resolution LiDAR DTM for Scratchbury. Map A focuses on the site, Map B depicts its topographical environs (© Environment Agency copyright and/or database right 2015. All rights reserved; Basemapping for Map B © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)

96

Multi-regional GIS-based Analysis of Hillfort Location and Morphology that surrounds Wheedlemont, and if defence was a primary concern during its construction, the site would have extended across the whole of the hilltop and would have had more substantial enclosing works (Shepherd 1996, 145). The limited defensive capability of Wheedlemont is also emphasised by the position of the entrance; this faces the north-eastern hillslope which forms a blind side to the hilltop as approaching traffic from these slopes is not visible. There is slight evidence for efforts having been made to defend Wheedlemont, this was noted by Jervise who identified a ditch on its southern and western aspects, which he argues were its most accessible areas (1871, 327). This demonstrates that although the site was not ostentatiously defended, efforts were made to counteract the susceptibility of the site from intruders, which was caused by its topographical location.

limited, enables analysis of physical and visual relations to have a concept of time. For example, at Hardings Down, three hillforts were situated on the same landform, and if contemporary it is likely that they were associated. Several of the test hillforts such as those in Warminster physically incorporate earlier monuments. The affect that this had upon the visual relations within this landscape potentially indicates the importance placed upon these monuments compared to the hillfort enclosure. These monuments likely initially drew people to these hilltops as they were already significant places. As this study develops, these potential associations are investigated through the visual and physical relationships that they exhibit. Physical relationship of the hillfort morphology and location with the landscape topography

Wheedlemont’s summit location influenced the degree to which the surrounding landscape was visible from it. Viewshed analysis from the hillfort found that, although a large proportion of the area that is within the 1km radius is visible, the area’s actual visual magnitude is low (Figure 79). The majority of the areas within the 3km radius are also visible, however beyond this radius, visibility becomes scattered. Within the 6km radius visibility is predominantly focused in the east, whilst within the 9km radius visibility becomes very scattered, but it is of high visibility.

As one would expect, a large number of the hillforts within this study were situated on the summit of hills, however the degree to which their morphology related to the topography was highly variable. Several of the hillforts were situated on domed hilltops that did not provide a readily definable platform for a site. These can be seen in both the Dartmoor and Aberdeenshire test areas where the morphologies of both Cranbrook Castle and Wheedlemont do not strictly follow the topographical form of the hilltop. Cranbrook is sited on the summit of a domed hill on the upland valley edge of the River Teign. The site does not strictly follow the shape of the hilltop which implies that its morphology was consciously selected to achieve a design that was enforced upon the landscape. The overall morphology also fails to respond to variations in the topography. For example, the banks which enclose the site are most numerous in the south which you would expect to be its most topographically susceptible aspect i.e. gently sloping, but it is not. Consequently, instead of responding to topographical weaknesses, this aspect of Cranbrook’s morphology appears to be a design enforced upon the landscape. According to Newman, the domed location of Cranbrook Castle meant that it has a 360°° vista (2011, 88). This is supported by the results of viewshed analysis from the hillfort grid (Figure 78). This demonstrates that on a whole landscape scale there is a 360° vista. However, the visibility of the different areas differs with distance, consequently at any given distance it is not 360°. The most visible aspects of the landscape are to the north and west, and the degree to which these areas are visible increases as distance from the site increases. The least visible areas are to the south, south-west, east and south-east.

Within the Warminster test area, Cley Hill is also situated on a domed shaped hilltop. Unlike Cranbrook and Wheedlemont, the morphology of Cley Hill largely concurs with the topography. The site’s enclosing works are situated downslope from the hilltop which physically accentuates the visual prominence of its interior. The topographical form of the hilltop also promotes intra-visibility across the site, for example from the northern and southern ends of the hillfort the majority of the enclosure is visible. The intra-site visibility is however restricted by the centrally positioned barrows. These features, particularly the larger of the two, are physically and visually prominent features within the main enclosure and from the wider landscape. Cley Hill’s topographic position promotes a high degree of visibility to the surrounding landscape. The results of viewshed analysis from the hillfort grid indicates that the site can see the majority of the area within the 1km radius; however, it is of a very low visual magnitude (Figure 80). Beyond this radius the visual magnitude of the visible areas increases, however the areas that can be seen become more scattered. The site has very limited visibility to the areas to the south of it. The most visible areas beyond the 1km radius are to the east and west of the site.

Moving to the Aberdeenshire test area, Wheedlemont is situated on the eastern summit of Wheedlemont hill. This domed hilltop is the most topographically defined aspect of the hill as the remainder is gently rolling. Although Wheedlemont is situated on the summit of the hill it does not occupy the entirety of this piece of land; neither does the morphology of the site differ with changes in topography. There is a substantial amount of wasted space

The two remaining sites in the Warminster test area are situated on the perimeter of Salisbury Plain. Although Scratchbury’s morphology is relatively uniform throughout, there is a slight differentiation in the height of the banks in relation to the topography. For example, the enclosing works on the eastern side of the site are 97

A GIS-based Analysis of Hillfort Location and Morphology

Figure 78 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Cranbrook Castle as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Dartmoor National Park Historic Environment Record 2013)

98

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 79 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Wheedlemont as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Aberdeenshire Council Archaeology Service 2014; Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

99

A GIS-based Analysis of Hillfort Location and Morphology

Figure 80 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Cley Hill as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)

higher and more prominent than those elsewhere on the circuit. This trait was also highlighted by McOmish and others in 2002 (75). Additionally, just like Cley Hill, the interior of the site is slightly domed and the bank and ditch which surround this enclosure, particularly on the western side are situated downslope. This meant that the

site had limited defensive capabilities and its interior was highly visible (Ibid.). The north-eastern and north-western entrances both have an in-turned terminal and are situated adjacent to highly sloping land. The south-eastern entrance was formed by a simple break in the rampart circuit. This is also positioned adjacent to the flatlands of the plateau. 100

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 81 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Scratchbury as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)

Scratchbury has wide ranging views to the 1km radius (Figure 81). Although the site has visual access to 360˚ around the site, these visible areas are of low visibility. As distance from the site increases, the visual magnitude of the visible areas also increases. Beyond the 3km radius the span of the visible areas becomes limited and it does not

cover a 360˚ vista. The visual magnitude of these visible areas is also highly variable. Located 2km to the north-west of Scratchbury is Battlesbury which is situated on a wide terminal of an upland outlier of Salisbury Plain. The site’s morphology, 101

A GIS-based Analysis of Hillfort Location and Morphology Cairnmore’s topographical position meant that it has scattered visibility to the surrounding landscape. Distance from the site does not affect visual magnitude, but it does affect what can be seen. The spread of visibility is very variable, but the majority of the site can see to the north and east. Between the hillfort and the 1km radius there is very scattered visibility to the north and south (Figure 83). Visibility is primarily focused to the northern and eastern areas between the 1km and 3km radii. Between the 3km and 6km radii visibility is more scattered to the north, east and west. Beyond these radii, visibility becomes very scattered to the north-east and south-west.

for the most part, adheres to the topography. However, the number of enclosing banks changes with variations in the topographical form of the land. For example, the number of enclosing banks is greatest at the northern end of the site, where it is adjacent to Battlesbury’s predecessor; Battlesbury Bowl settlement. The extensive enclosure of this northern sector reinforced the strength of the enclosure which implies that attempts were made to counteract the susceptibility of this area to intruders, equally it may also be evidence for them physically disassociating themselves from their predecessor. Multiple enclosing works were also employed on the southern corner of the site. This is an area that was naturally defended and defined by the highly sloping topography, therefore such extensive enclosure was not needed. Forming extensive enclosing works in an area where it was not defensively needed implies that defence was not the sole motivator whilst investing in this site. This implies that other morphological components such as the northern enclosing works may have been motivated by other factors such as symbolism and display.

The two remaining sites in the Aberdeenshire test area comprise a summit located vitrified inner enclosure and a downslope outer stone banked enclosure. Dunnideer, crowns the summit of the southwestern hilltop of a southwest-northeast hill range. The hilltop is well defined, and the morphology of the inner and outer forts distinctively follow the line of the various topographical stages of the hill. The vitrified inner fort is oblong and encloses the summit of the hill, whilst the outer fort adheres to the perimeter of the hillslope. Consequently, it does not form a visually prominent area. The enclosure of this area also failed to substantially increase the amount of usable space at this site as there are very few areas within this outer fort that are easily habitable due to the slope of the land. The topographical form of this hilltop allowed for a site of this morphology to be constructed, however as Feachem states, it did not restrict the site to this formation; other forms were possible (1966, 68). The sequence of the inner and outer fort is not known, but it is thought to represent at least two phases of occupation within the hillfort (Shepherd 1996).

The placement of entrances at Battlesbury was also seemingly more complex and motivated by more factors other than simply enabling access. The site has two entrances, one in the north-west and the other in the east. The north-western entrance is situated on a straight portion of the hillfort’s enclosing works and it opens out onto an area of steep slopes. If it was positioned a matter of metres to the east, it would have faced the plateau and enabled access. However, this plateau also housed Battlesbury’s predecessor, the Bowl settlement, which, although it contracted with the establishment of Battlesbury, efforts may still have been made to disassociate the two. Contrastingly, the eastern entrance is situated on a corner of the site towards a relatively flat aspect of the hilltop which enabled access.

In terms of the visibility of the wider landscape from this site, Dunnideer’s location promotes a high degree of visual accessibility to the east from the entire site. However, the outer fort has a greater degree of visual accessibility to the area within the 1km radius compared to the inner fort (Figure 84). Beyond this radius, values of visual magnitude increase with the highest values occurring beyond the 3km and 6km radii in the north, south and east.

The degree to which the physical relationship between the morphology and the topography at Battlesbury promoted interaction between those inside and outside the hillfort was variable, as to was the visibility of the landscape from the site. Battlesbury’s location promotes a high degree of visibility to the areas within a 3km radius of the site (Figure 82). Towards the 6km and 9km radii, visibility becomes increasingly scattered, the western and southern aspects of the landscape are also predominantly visible.

Unlike Dunnideer, the vitrified inner fort at Tap o’Noth is positioned slightly offset from the summit. This further supports the observations made by Feacham at Dunnideer, that whilst this location enabled such a morphology, it did not dictate it. The addition of the oblong inner fort was a design which was applied to this hilltop. The outer bank generally follows the shape of the hill, consequently the shape of the hillside defined the shape of the fort. This bank is less substantial than the inner bank.

Moving back to the Aberdeenshire test area, evidence indicates that Cairnmore’s morphology is oriented to respond to topographically weak points. The enclosure is substantially enhanced along the eastern and western sides where the slope of the land is least. On the western side, the banks and ditches are spaced closely together, and these earthworks cut the hillfort off from the adjacent western spur of land. On the eastern side, the gap between the inner and middle ramparts is greater than on the western side; this enhances the entranceway as it physically lengthens the entrance passage. It also acts as a further means to define and potentially make use of space.

The overall topographic position of Tap o’Noth enables the site to have a good degree of visual access to a large proportion of its environs, particularly the areas further afield. For example, although there is a wide degree of the 1km radius that is visible, the visible areas have a visual magnitude that generally falls within the low visibility 102

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 82 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Battlesbury as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)

103

A GIS-based Analysis of Hillfort Location and Morphology

Figure 83 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Cairnmore as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Aberdeenshire Council Archaeology Service 2014; Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

104

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 84 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Dunnideer as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM, Map A depicts the visibility from the Inner Fort and Map B depicts it from the Outer fort (Aberdeenshire Council Archaeology Service 2014; Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

105

A GIS-based Analysis of Hillfort Location and Morphology

Figure 85 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Tap o’Noth as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM, Map A depicts the visibility from the Inner Fort and Map B depicts it from the Outer fort (Aberdeenshire Council Archaeology Service 2014; Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

106

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 86 View from of Prestonbury from Broadmoor Common (Jessica Murray 2014)

overlooks the site. Although visibility into the site is not hindered from the east, the number of enclosing works within this area indicates that its morphology responds to the topography in that it forms an artificial enclosure within an area that the site is not naturally defined. However, the failure to create an ostentatious enclosure meant that this site was visually open, and the openness was accentuated by its hillslope location.

category (Figure 85). The visual magnitude of the visible areas increases towards the 3km radius; however both the inner and outer fort fail to have visual accessibility to the north-east/east whereas the remainder of the landscape is visible. The most visible area is the west. Beyond the 3km radius the visibility of the surrounding landscape becomes more scattered. The outer fort has a greater degree of visual accessibility to the surrounding landscape as the areas to the west have a higher visual magnitude.

Prestonbury’s hillslope location influenced its visual accessibility of the wider landscape. The GIS-based viewshed analysis from the hillfort grid demonstrates that the majority of the 1km buffer zone is visible from the site (Figure 87). The majority of the northern area of the 3km radius is visible with only scatterings of the east and west, and areas to the south are not visible. Visibility decreases with distance in relation to the 6km and 9km radii with only scattered areas of the north being visible in the 6km radius and in the 9km radius only a limited area to the west and north-east are visible.

The second most abundant topographic relationship within this study was the hillslope enclosure which occurred in both the Dartmoor and Gower test areas. By their very nature hillslope enclosures were highly directional and this caused a high degree of visual accessibility into their interiors. Prestonbury Castle is located on the edge of the Teign river valley. The degree of artificial enclosure decreases on the western valley side where only the inner enclosure bank joins the valley edge. On Prestonbury’s eastern side, the perimeter bank is the most physically prominent enclosing work; this faces Broadmoor Common which overlooks the site. Although this bank is an enhanced enclosing work, it does not restrict visibility into the interior. This is due to the topography of the land which slopes up towards the west. Positioning a timber palisade on top of any of the banks would also have failed to significantly hinder visibility into the site.

Wooston Castle is also located on a spur of the Teign Valley, and the principal shape of the main enclosure is defined by the topography, although the site’s outer banks do not adhere to it. The inner most bank and ditch are the site’s most prominent aspect and form the southern perimeter to the main inner enclosure. The site’s outworks face uphill, as does the hollow way which lines the route into the site. Quinnel and others highlighted that whilst the term hollow way implied that it was a track that had been worn into the landscape over time, due to the depth of the feature it is likely that it was partially dug out in prehistory (2013, 12).

This was illustrated through the results of the site visit; Figure 86 illustrates that the majority of the interior is visible from this point on Broadmoor Common which 107

A GIS-based Analysis of Hillfort Location and Morphology

Figure 87 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Prestonbury Castle as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Dartmoor National Park Historic Environment Record 2013)

108

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 88 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Wooston Castle as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Dartmoor National Park Historic Environment Record 2013)

109

A GIS-based Analysis of Hillfort Location and Morphology

Figure 89 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from The Bulwark as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014))

1km radius as the majority of this area, apart from areas to the south-east, is visible. Visibility is restricted beyond the 1km radius, with only scattered areas of the northern and western area of the 3km radius, the northern area of the 6km radius and a very small aspect of the eastern area within the 9km radius are visible.

Whilst the topographic position of the site meant that it was overlooked from the south, it also hindered the site’s visibility of the surrounding landscape which is of very limited visibility (Figure 88). Visibility is primarily restricted to areas to the north with intermittent visibility to the east and west. The site has greatest visibility to the 110

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 90 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Hardings Down West Enclosure as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014))

On the Gower, the morphology of the hillslope enclosure; The Bulwark, strongly adheres to the topographical form of the hillside that it is situated on. The site’s outer eastern earthworks extend out towards the eastern edge of this hilltop. This extension did not enlarge the site’s habitation area as there is no visible evidence for occupation here.

Instead, the extension enhanced the site’s entrance passageway. This implies that to the hillfort builders it was important to secure the eastern edge of Llanmadoc Hill. Further motivations that went beyond the practical need of enclosure was indicated by the heavy enclosure of the site’s southern aspect which was not needed on defensive 111

A GIS-based Analysis of Hillfort Location and Morphology

Figure 91 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Hardings Down North Enclosure as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014))

grounds due to the highly sloping nature of the topography.

Visibility was also restricted to the remainder of Llanmadoc Hill. The site has a greater degree of visibility to the land further away from the site, particularly to the east.

Due to The Bulwark’s hillslope position it does not have 360˚ views to the surrounding landscape, neither is there complete intra-visibility across the enclosure (Figure 89).

To the south on Hardings Down, although the West Enclosure is also a hillslope enclosure, it largely fails to 112

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 92 Viewshed results from the hillfort grid, depicting the visibility of the surrounding landscape from Hardings Down East Enclosure as defined by the hillfort buffers. Viewshed analysis was based upon the Ordnance Survey DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014))

adhere to the topographical form of the hillside as the enclosing works cut across the hillside. It does however respond to the topographical weaknesses of this location. For example, in the southern corner of the site, where the site faces the summit of Hardings Down and the East Enclosure; the banks are most numerous with a main,

middle and outer rampart. As with many of the hillslope enclosures this topographical location reduced the site’s ability to see the surrounding landscape as it was impossible for it to have 360˚ views (Figure 90). This is demonstrated by the fact that the site had very restricted visibility to the south. Visual accessibility within the 3km 113

A GIS-based Analysis of Hillfort Location and Morphology beyond a functional need and were potentially motivated by ideas of display which will be investigated below.

radius is relatively low, and predominantly confined to the north, west and south. Beyond this radius the visual magnitude of the visible areas increases; these visible areas are predominantly confined to the coastal areas in the north-west.

Whilst topography, in many cases, influenced the morphology of these test hillforts, their visual prominence was also heavily influenced by the topography. Erosion has however strongly affected the visual prominence of the enclosing works, therefore what is seen today would not have been the case when the hillforts were in use. However, at several sites such as Tap o’Noth and Wheedlemont their enclosing works are not visually prominent. Even though erosion is likely to have influenced their visual prominence it is unlikely that the outer bank at Tap o’Noth for example was as prominent as the inner fort bank or the enclosing works at sites such as Battlesbury. At hillforts where the enclosing works were of limited visual prominence the topographical prominence of the location enhanced the prominence of the sites.

Unlike the West Enclosure, the North Enclosure’s morphology is uniform. The enclosure slopes down to the north, which means that the site is highly visible from this direction. This sloping topography affects the visibility of the landscape from the site, as it has a higher degree of visual accessibility to the north than to the south (Figure 91). The morphology of Hardings Down East enclosure is relatively uniform, apart from in the east, where there is a series of outworks that are associated with the entrance. These outworks face towards the gentle eastern slopes of Hardings Down. They also act as an extension to the footprint of the site by taking in the break of slope. Even though the site fails to enclose the summit, it is positioned sufficiently close to it that it enables the site to have a large degree of visual accessibility to the surrounding landscape. This visual accessibility is particularly high to the 1km radius in the north/north-east and the south-west (Figure 92). A large proportion of the area between the 1km and 3km radii is visible. Between the 3km and 6km radii, visibility is confined to the north-west and east. The most visible area is the east, this has a visual magnitude within the upper percentile range. Visibility is restricted to the coastal areas in the north-east and west between the 6km and 9km radius.

Topography also strongly influenced the visual accessibility of the surrounding landscape from the hillforts. The variability in this effect is strongly exhibited at two distinct topographical locations: the domed hilltop and the hillslope. Sites that are situated on domed hilltops such as Cranbrook Castle and Wheedlemont. largely promote 360-degree views out towards the surrounding landscape. Whilst hillslope enclosures such as the Hardings Down Group have inhibited visibility towards the surrounding landscape that are highly directional. Here, the variability of the relationship between the underlying topography and hillforts were explored, alongside the affect that this has upon their visual relationships. Below the effect of topography on the visual and physical prominence of these sites alongside the image that they portrayed is explored further.

Summary The variability of the test areas was further supported by the variability of their topographical relationships. Sites range from those that enclose well defined topographical locations; such as Cley Hill, Tap o’Noth and Dunnideer, to those that were located in less well defined locations; like Cranbrook Castle and Prestonbury Castle. In these instances, designs were applied upon the landscape and other morphological forms were possible. For example, the inner vitrified wall at Tap o’ Noth was positioned offset from the summit when in fact it could have occupied the entire summit.

Image Whilst the relationship of hillfort morphology with topography was highly variable, the degree to which hillforts were visible from the wider landscape and consequently the images that they portrayed was also variable. These images were informed by the differing degrees of visual accessibility of the architectural components and the underlying relationship between the hillfort and the topography. Visual accessibility was a platform by which information could be accessed by those which were not necessarily part of the same community that built these hillforts. The accessibility of this information varied across test areas and through moving through and being within these landscapes. The degree to which these sites were completely understood on a morphological basis was highly variable. In some instances, it was not possible to see all of their enclosing works, whilst at others it was.

As expected, the majority of the hillforts’ enclosing works were positioned to respond to the topographical weaknesses of these locations. For example, at Battlesbury one of its most enclosed aspects faced the landward approach and the entrance was oriented offset from it. However, at Prestonbury although the most enclosed aspect of the site faced the landward approach, these enclosing works did not inhibit visibility, they formed an elaborate demarcation of space. At other sites such as The Bulwark their morphology was ostentatious and went beyond satisfying a defensive need. For example, its southern enclosing works were not defensively needed due to the slope of the topography, these earthworks went

For example, cost surface analysis found that Prestonbury portrayed a relatively complete image to the surrounding landscape as there are several places where there is 114

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 93 Results of cost surface analysis highlighting the routes where Prestonbury’s enclosing works are most visible (Dartmoor National Park Historic Environment Record 2013; © Environment Agency copyright and/or database right 2015. All rights reserved. copyright and/or database right 2015. All rights reserved)

image to the wider landscape. The site is most visible from the 1km radius, with a visual magnitude between the 50th and 74th percentile (Figure 96). However, field photography indicated that from Viewpoint 1 which is to the south-east of the site there was no visual accessibility into it (Figure 97). From within the 3km radius at Viewpoint 4, which is to the north-north-east of the site, the northern banks are visible but they are not visually prominent. Within the 9km radius at Viewpoint 12 the site’s peak is clearly visible, however it is difficult to distinguish its morphological components.

maximum visibility to its architectural components (Figure 93). The opportunities to see the extent of these enclosing works is integral to the visual integration of this site within the landscape. The relatively complete image of Prestonbury from Wooston Castle is also supported by the viewshed results from this site as the majority of its southern and eastern aspects are visible (Figure 94). Unlike the results from Wooston Castle, Prestonbury’s complete image which was implied by the viewpaths was not supported by the viewshed results from Cranbrook Castle (Figure 95). These indicate that the western and southern aspects of Prestonbury are the only visible parts of the site, and they are of low visibility. Visual access to the enclosing works is therefore only possible when moving through the landscape.

Another hillfort whose visual qualities enables the site to portray a relatively complete image is Battlesbury. Cost surface analysis demonstrated that whilst travelling between Battlesbury, Cley Hill and Scratchbury it is possible to gain a largely complete image of the site (Figure 98).

According to the viewshed results from the surrounding radii, Prestonbury Castle also exhibits a relatively complete 115

A GIS-based Analysis of Hillfort Location and Morphology

Figure 94 Results of viewshed analysis indicating the visibility of Prestonbury Castle from Wooston Castle, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR(DTM for viewshed analysis in Map A © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence), © Environment Agency copyright and/or database right 2015. All rights reserved.)

fieldwork photography illustrates a variable picture. For example, from the 1km radius at Viewpoint 21, which is to the west of the site, Battlesbury’s enclosing works are visually prominent (Figure 104). From this point the northwestern enclosing banks are clearly visible and visually dominant. The north-western entrance is also visible; however, its interior is not. However, from the 3km radius at Viewpoints 32 and 1, Battlesbury’s interior is not visible but its enclosing works are visually prominent.

This site’s ability to portray a complete image towards its environs is reflected in both the results of the viewshed analysis and fieldwork observations. The relatively uniform nature of Battlesbury’s enclosing works affords its complete image. The viewshed analysis from Cley Hill indicates that the western side of Battlesbury is of high visibility (Figure 99) Although the viewshed results indicate that this aspect of the site is of high visibility, the field photography illustrates that the enclosing works are not discernible due to the distance between the sites (Figure 100). Similarly, the viewshed results from Scratchbury indicate that at least one section of each of the enclosing works were visible from the outset and consequently support the results of the cost surface analysis. Visibility is, however, restricted to the south-eastern end of the site (Figure 101). Fieldwork photography illustrates that the enclosing works are not as visible as the interior, but they are visually prominent (Figure 102).

Whilst Battlesbury and Prestonbury exhibit a relatively complete image, there are a number of other hillforts that did not. In the Gower, The Bulwark’s image is split in two whereby the outermost works which comprise the outer bank, the outer ditch and the eastern entrance’s outworks are most visible from the south-west whilst the remaining architectural components are most visible from the south. These routeways do not follow the line of any distinct topographical or archaeological features (Figure 105).

The visibility of Battlesbury’s enclosing works is supported by the viewshed results from the surrounding radii. These results indicate that the enclosing works are most visible from the 3km radius as too is its interior (Figure 103). The

Although cost surface analysis illustrates that the visual qualities of The Bulwark’s architectural components were split in two, the viewshed results from Hardings Down 116

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 95 Results of viewshed analysis indicating the visibility of Prestonbury Castle from Cranbrook Castle, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

indicate that only the site’s southern sector is visible (Figure 106-108). This comprises the majority of the enclosing works. The photography from Hardings Down supports the viewshed results as its southern aspect is the only visible component (Figure 109). Although this sector is visible in the photography, the enclosing works are neither clearly distinguishable nor visually prominent.

image, cost surface and viewshed analyses demonstrate that Cley Hill portrays a very fragmented image. It was not possible to travel towards Cley Hill from Battlesbury and Scratchbury and to maintain a complete image of the site (Figure 112). The viewshed results from Battlesbury (Figure 113) and Scratchbury (Figure 114) indicate that its eastern aspect is in fact visible from them and this area comprises both the inner and outer bank. A static viewpoint from these sites would therefore have access to these enclosing works however movement through the landscape and its undulating topography inevitably effected their visibility.

The Bulwark’s visibility is much more varied from the wider landscape; however, it is most visible from the 6km radius as the majority of the site is of high visibility (Figure 110). The site’s variable visibility from the surrounding landscape is supported by the fieldwork photography. From Viewpoint 4, which is to the north-west of the site, Llanmadoc Hill is visually imposing (Figure 111), the outline of the site’s northern half is also visible, but it is not clearly distinguishable. The Bulwark’s enclosing works are also visible from Viewpoint 5 which is to the northwest of the site. Viewpoint 2 which is to the south-west of it also has visibility of The Bulwark’s enclosing works.

Cley Hill is of greater visibility from the wider landscape and its visibility increases with an increase in distance from the site. It is most visible from the 6km and 9km radii whereas it is of scattered visibility from the 1km radius (Figure 115). This is supported by the site photography (Figure 116). The fragmentary nature of the image portrayed by Cley Hill is caused by the downslope position of the enclosing works which form one of the most visually imposing aspects of the site. The visual prominence of the hillfort’s eastern aspect is illustrated from Viewpoint 13

Unlike Prestonbury, Battlesbury and The Bulwark whose architectural components form a relatively coherent 117

A GIS-based Analysis of Hillfort Location and Morphology

Figure 96 Results of viewshed analysis from the radii towards Prestonbury depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

118

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 97 Visibility of Prestonbury Castle from the surrounding landscape. Map A depicts the view from Viewpoint 1, Map B depicts the view from Viewpoint 4, Map C depicts the view from Viewpoint 12 whilst Map D illustrates the location of the Viewpoints. (Dartmoor National Park Historic Environment Record 2013; Jessica Murray 2013; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

119

A GIS-based Analysis of Hillfort Location and Morphology

Figure 98 Results of cost surface analysis highlighting the routes where Battlesbury’s enclosing works are most visible (© Environment Agency copyright and/or database right 2015. All rights reserved.; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018))

which is to the south-west of the site. Although the site’s eastern aspect is visually dominant, the actual hillfort enclosure is not the most visually imposing aspect, instead this is the topographical landform. Viewpoint 16 has visual accessibility to the site’s enclosing works and the central barrows. From the 6km radius at Viewpoint 24, Cley Hill’s eastern side and the central barrows are visually prominent. However, the distance between this location and the site meant that visual clarity was poor and the enclosing works were not easily discernible.

For example, from the 1km radius, at Viewpoint 2, the site is of limited visibility and visual clarity. Here, only the south-eastern enclosing works are visible, but they are not visually prominent. Just as from Viewpoint 2, the visual clarity of Cranbrook Castle from Viewpoint 3, which is 2.1km to the south-east, remains poor. From the 3km at Viewpoint 4 the site’s enclosing works are not visible but its interior is visible. Viewpoint 7 is situated 4.6km to the north-west of Cranbrook. From this location the impact of Cranbrook on the landscape is clear (as the hilltop has been levelled and the western enclosing works are visible.

Another site which is most visible from long distances is Cranbrook Castle. The majority of Cranbrook is highly visible from the 9km radius (Figure 117), however the field photography illustrates a variable picture (Figure 118).

Just as Cranbrook’s visibility differs with distance, it also changes from the neighbouring hillforts. For example, 120

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 99 Results of viewshed analysis indicating the visibility of Battlesbury from Cley Hill, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Figure 100 View of Battlesbury from Cley Hill (Jessica Murray 2013)

121

A GIS-based Analysis of Hillfort Location and Morphology

Figure 101 Results of viewshed analysis indicating the visibility of Battlesbury from Scratchbury, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Figure 102 View of Battlesbury from Scratchbury (Jessica Murray 2013)

122

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 103 Results of viewshed analysis from the radii towards Battlesbury depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

123

A GIS-based Analysis of Hillfort Location and Morphology

Figure 104 Visibility of Battlesbury from the surrounding landscape. Map A depicts the view from Viewpoint 21, Map B depicts the view from Viewpoint 32 Map C depicts the view from Viewpoint 1 whilst Map D illustrates the location of the Viewpoints. (Jessica Murray 2013; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)))

124

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 105 Results of cost surface analysis highlighting the routes where The Bulwark’s enclosing works are most visible (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; information held by the GGAT HER Charitable Trust copyright (provided 2014))

125

A GIS-based Analysis of Hillfort Location and Morphology

Figure 106 Results of viewshed analysis indicating the visibility of The Bulwark from Hardings Down North Enclosure, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 2m resolution LiDAR (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

although Prestonbury is only 1.6km away from Cranbrook Castle, Cranbrook is of limited visibility. Viewshed analysis indicates that only the northern and northeastern aspects of Cranbrook Castle are visible (Figure 119). Viewshed analysis from the 1m resolution LiDAR clearly indicates that the northern and eastern banks are one of the site’s most visible aspects. The similar visual characteristics of these enclosing works are supported by the results of the cost surface analysis (Figure 120). This identified that although the visibility of Cranbrook was fragmentary when travelling from Prestonbury, the visual qualities of the northern and outer banks are very similar as there are large areas on the route where both are visible. It was difficult to confirm the results from viewshed analysis in the field as the northern area of Cranbrook is its least extant, consequently it is not distinguishable through photography. The similar visual qualities of the northern and outer banks at Cranbrook are also echoed in the viewshed and cost surface analysis results from Wooston Castle. Although Cranbrook Castle is of limited visibility from Wooston, the north-eastern corner of the

site is visible (Figure 121). This is also supported by the cost surface analysis which highlights that the northern and outer enclosing works have similar visual qualities as there are long stretches on this route that have visual access to these components. Whilst sites such as Cranbrook have far reaching audiences, the hillforts of the Aberdeenshire test area have wider audiences that span both the immediate and long-distance environs. In particular, Wheedlemont is highly visible from all of the surrounding radii (Figure 122). Similarly, the viewshed analysis and fieldwork photography demonstrate that the site is of high visibility from Cairnmore which is situated 3.3km to the south-east (Figure 123-124). Although the site is of good visibility from Cairnmore, it is not visually prominent as the enclosing works are poorly distinguishable, the topographical location in itself is, however, visually prominent. Fieldwork photography indicated that this location is also visually prominent from Tap o’Noth, but mist on the day meant that visual clarity was poor (Figure 125). The viewshed results indicate that 126

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 107 Results of viewshed analysis indicating the visibility of The Bulwark from Hardings Down West Enclosure, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 2m resolution LiDAR (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

6km radius as the majority of the site is of high visibility. The visibility of the site decreases slightly from the 9km radius. Although Cairnmore is generally highly visible from the surrounding landscape, it is poorly visible from the neighbouring hillforts. This poor visibility does not reflect the distance between the sites as they all have poor visual accessibility of the site regardless of their distance from Cairnmore. For example, from Wheedlemont, which is closest to Cairnmore, the majority of the site is visible, but it is of low visibility (Figure 130). This is supported by the field photography, where although the site is visible, it is impossible to distinguish the enclosing works (Figure 131). Cairnmore is of greater visibility from Tap o’Noth (Figure 132). There is limited visibility to Cairnmore from Dunnideer (Figure 133) as only the site’s north-eastern aspect is visible with a low value of visual magnitude. Dunnideer exhibits similar visual qualities to Cairnmore, it is most visible from the 6km radius, where the majority of the site has a visual magnitude within the upper quartile range (Figure 134). The field photography illustrates that there is an increase in the location’s visual prominence

the site is of moderate visibility from Tap o’Noth (Figure 126). Wheedlemont is poorly visible from Dunnideer (Figure 127). Wheedlemont’s summit position meant that it was visually and physically prominent within the surrounding landscape (Figure 128). However, due to the site’s poor survival it is very difficult to distinguish how its prominence related to the topography or its morphology. Whilst approaching the site from the lower ground the enclosing works are not visible, although this is unlikely to have been the case when the site was in use as they have inevitably eroded over the centuries and they are likely to have been substantially higher. The site’s enclosing bank sits downslope from the interior consequently the interior was likely to be visible from the surrounding landscape, to a certain degree. Also in the Aberdeenshire test area, Cairnmore is of relatively high visibility from the 3km, 6km and 9km radii (Figure 129). The site’s most visible aspect from the 3km radius is its northern half, which is of high visibility. The site’s visual magnitude is greater from the 127

A GIS-based Analysis of Hillfort Location and Morphology

Figure 108 Results of viewshed analysis indicating the visibility of The Bulwark from Hardings Down East Enclosure, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 2m resolution LiDAR (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

supported by the fieldwork photography. From Viewpoint 3, the enclosing works are of relatively clear visibility but the site’s interior is not (Figure 140). The field photography from the 3km radius demonstrates that the site is highly visible and visually prominent. Although Tap o’Noth is visually prominent, visual clarity is poor, and any interior activity could only be seen from within the outer fort as the inner fort bank obstructs visibility from the surrounding lowlands. Although the inner bank obstructs visibility into the inner fort, it is visually and physically prominent: this prominence is accentuated by its summit position. Field photography demonstrates that from Viewpoint 10 it looks as though the inner fort is positioned on the summit of the hill, however, from Viewpoint 12 the actual offset position of the bank is visible. The visibility of the site decreases slightly from the 9km radius. From this distance the visual magnitude of the northern aspect of the site decreases, whilst the remainder of the site is of high visibility.

(Figure 135) with an increase in distance. Although the topographical land form is visually prominent, the site’s enclosing works remain poorly distinguishable. It was therefore not possible to clarify how changes in direction effect the ‘image’ of the site. The visibility of Dunnideer from the surrounding hillforts is variable. From Tap o’Noth the site’s north-western, western and south-western aspects are of moderate visibility (Figure 136). The western and southern aspects of Dunnideer are visible to varying degrees from Wheedlemont and the values of visual magnitude are highly variable, ranging from low to moderate visibility (Figure 137). It is poorly visible from Cairnmore as only a small section of the southern half of the inner fort is visible (Figure 138). Moving to the second vitrified fort in the Aberdeenshire test area, Tap o’Noth, is of highly variable visibility but it is most visible from both the 3km and 6km radii (Figure 139). The variable visibility of the site’s different components is

Whilst the viewshed results from the radii indicate that the majority of Tap o’ Noth is visible from the surrounding 128

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 109 View of The Bulwark from Hardings Down (Jessica Murray 2013)

landscape, this is not the case from the surrounding hillforts. Dunnideer is believed to have been in use at the same time that the inner fort at Tap o’Noth was constructed (Cook 2013). However, this interpretation was only based upon a comparison with the dated and morphologically similar inner fort at Dunnideer. Viewshed analysis from both the inner and outer fort at Dunnideer indicates that the eastern sides of both the inner and outer forts at Tap o’Noth are of moderate visibility (Figure 141). This is supported by the field photography from the inner fort. From this point, Tap o’Noth and the Hill of Noth are visually prominent, however the distance between the sites is great, and consequently visual clarity is poor (Figure 142).

Whilst there were a number of sites that had very large- scale audiences, there are a several others whose audiences are much smaller and this is demonstrated within the Gower test area. For example, Hardings Down North Enclosure is most visible from the 1km radius as its visual magnitude is within the upper quartile range (Figure 147). The high visibility of this enclosure from within the 1km radius is supported by the fieldwork photography from Viewpoints 2 and 3 (Figure 148). These viewpoints are to the north-west of the site and both have total visual accessibility to it. The site appears well defined, but the enclosing works are not imposing. Although the North Enclosure is of relatively good visibility from its wider environs, it is of poor visibility from the neighbouring hillforts, especially those on Hardings Down (Figures 149150). The viewshed results indicate that the site is more visible from The Bulwark than from the sites on Hardings Down (Figures 151). The complete visual accessibility of the North Enclosure from The Bulwark is supported by the fieldwork photography (Figure 152). This illustrates that the whole enclosure including its enclosing works are visible, however it is not visually dominant.

Unsurprisingly, due to the distance between the sites, Wheedlemont and Cairnmore have much greater visibility of Tap o’ Noth. Both of these sites have a high degree of visual accessibility to the southern aspect of the outer fort, and also to the majority of the vitrified inner wall (Figures 143-144). The high degree of visibility to Tap o’Noth from these sites is supported by fieldwork photography. However, it also demonstrates that the height difference and the distance between the sites meant that although the site is visually prominent, visual clarity is poor (Figures 145-146).

Hardings Down East Enclosure has a slightly wider audience than the North Enclosure as it is most visible 129

A GIS-based Analysis of Hillfort Location and Morphology

Figure 110 Results of viewshed analysis from the radii towards The Bulwark depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (DTM for analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

130

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 111 View of The Bulwark from the surrounding landscape. Map A depicts the view from Viewpoint 4, Map B depicts the view from Viewpoint 2, Map C depicts the views from Viewpoint 5 whilst Map D illustrates the location of the Viewpoints (Jessica Murray 2013; Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014); Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

131

A GIS-based Analysis of Hillfort Location and Morphology

Figure 112 Results of cost surface analysis highlighting the routes where Cley Hill’s enclosing works are most visible (© Environment Agency copyright and/or database right 2015. All rights reserved.; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)

landscape (Figures 158). None of the pathways, which are based upon the visual accessibility of the site’s individual components, follow the same route. The variability in the visual qualities of the East Enclosure’s enclosing works is supported by the results of viewshed analysis from the surrounding radii as they differ greatly in visual magnitude.

from the 1km and 3km radii (Figure 153). There is limited visibility between this site and the other hillforts on Harding’s Down. It is not visible from the North Enclosure (Figure 154), and only the very outskirts of the site are visible from the West Enclosure (Figure 155). The site’s visibility increases from The Bulwark as the majority of the East Enclosure is visible (Figure 156). The field photography from The Bulwark supports the viewshed results (Figure 157). However, the site is not clearly distinguishable, today it can only be visually referenced by its summit location and the banks that cut across the hilltop.

As a whole Hardings Down West Enclosure is most visible from the 1km and 9km radii (Figure 159). Field photography indicates that the majority of the site’s interior is visible from Viewpoints 3 and 2 which are situated to the north-east of the site (Figure 160). Whilst the majority of the West Enclosure is to varying degrees visible from the surrounding landscape, it is of limited visibility from

Cost surface analysis found that the East enclosure portrays a very fragmented image out to the surrounding 132

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 113 Results of viewshed analysis indicating the visibility of Cley Hill from Battlesbury, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Conversely, within the Dartmoor test area, the visibility of Wooston Castle is highly restricted and the audience is limited. From the 1km radius there is visibility to the entirety of the site to varying degrees (Figure 166). Its western aspect and the hollow way are its most visible aspects as they have a visual magnitude between the 50th and 74th percentile, whilst the remainder of the site is of low visibility with a value within the lower quartile range. The overall visibility of Wooston Castle decreases from the 3km radius. The site’s western aspect is largely not visible; the western areas that are visible are of low visibility. The remainder of the site is of variable visibility, this has values of visual magnitude ranging from the 25th percentile to the 74th. Its most visible aspects are the southern terminus of the hollow way and the south-eastern area of the main hillfort enclosure. There is limited change in the visibility of Wooston Castle between the 3km and 6km radii. The site’s western aspect is not visible, whilst the southern terminus of the hollow way remains to be one of the most visible parts. The remainder of the site is

its neighbours on Hardings Down. The outer southeastern enclosing works face the East enclosure; these obstruct visual accessibility to the inner enclosure (Figure 161). Viewshed analysis found that these earthworks are among the few areas of the site that are visible from the East Enclosure but are of low visibility. This was also confirmed by the field photography (Figure 162). From the North enclosure, the West enclosure is of limited visibility (Figure 163). The majority of the visible areas are confined to the outskirts of the site; these include a large area of the annexe. These areas have highly variable values of visual magnitude. Unlike from the sites on Hardings Down, the majority of the West enclosure is visible from The Bulwark but it is of low visibility (Figure 164). Whilst the site is of varying visibility from the neighbouring hillforts, cost surface analysis found that it portrayed a fragmentary image. It is not possible for people to gain a complete visual understanding of the site’s morphology when travelling towards the site from neighbouring hillforts (Figure 165). 133

A GIS-based Analysis of Hillfort Location and Morphology

Figure 114 Results of viewshed analysis indicating the visibility of Cley Hill from Scratchbury, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

of relatively low visibility with visual magnitude values ranging from the 1st and 49th percentile. Visibility continues to decrease from the 9km radius where the visible areas decrease and the overall visual magnitude of these areas decreases.

indicate that the visible area is confined to the western portion of the site (including the hollow way); however, the higher resolution DTM indicates that visibility into the site is restricted. The fragmentary nature of Wooston Castle’s visual characteristics is also illustrated by the cost surface analysis which shows that there is not one location between Wooston Castle and the neighbouring hillforts where the site portrays a ‘complete’ image (Figure 169). The hollow way is mostly visible from the northern valley side of the Teign river valley, whereas the remainder of the site is visible from the land to the south side. The paths which have maximum visibility to the hollow way also interact with later prehistoric enclosures (I).

The visibility of the site from the neighbouring hillforts is also highly variable. From Cranbrook Castle, Wooston Castle is of limited visibility (Figure 167). The Ordnance Survey PROFILE DTM indicates that only the southern aspect of the hollow way and the northern corner of the site are visible with a visual magnitude within the lower quartile range. The 1m DTM also indicates that the site is of low visibility, but these results imply that the entire hollow way and the western bank are visible. The visual magnitude of the visible areas increases from Prestonbury (Figure 168) with the most visible aspect being the southern portion of the hollow way which is of high visibility, this is largely supported by both DTM friction surfaces. The remainder of the visible areas have a visual magnitude between the 25th and 74th percentile. Both of the DTM’s

Summary The variability of the evidence exhibited at the hillforts within these test areas is emphasised by their varying degrees of visual accessibility. As it was highlighted, it 134

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 115 Results of viewshed analysis from the radii towards Cley Hill depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

135

A GIS-based Analysis of Hillfort Location and Morphology

Figure 116 View of Cley Hill from the surrounding landscape. Map A depicts the view from Viewpoint 16, Map B depicts the view from Viewpoint 13, Map C depicts the views from Viewpoint 24 whilst Map D illustrates the location of the Viewpoints (Jessica Murray 2013; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)

136

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 117 Results of viewshed analysis from the radii towards Cranbrook Castle depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence)

137

A GIS-based Analysis of Hillfort Location and Morphology

Figure 118 View of Cranbrook Castle from the surrounding landscape. Map A depicts the view from Viewpoint 2, Map B depicts the view from Viewpoint 3, Map C depicts the view from Viewpoint 4, Map D depicts the view from Viewpoint 7, whilst Map E illustrates the location of the Viewpoints (Dartmoor National Park Historic Environment Record 2013;Jessica Murray 2013; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

138

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 119 Results of viewshed analysis indicating the visibility of Cranbrook Castle from Prestonbury Castle, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

is not possible to gain a complete visual understanding of the morphological composition of several sites within these test areas whilst moving through the landscape. For example, at The Bulwark cost surface analysis indicates that the inner and outer enclosures have variable visual characteristics therefore it is not possible to see them both at once. However, from a static position on Hardings Down, viewshed analysis indicates that the southern sector of both these enclosures can be seen. Alternatively, Prestonbury and Battlesbury both exhibit complete images to the surrounding landscape whereby there are opportunities to see the true extent of their enclosing works when moving through the landscape.

Just as the visibility of hillforts varied from the wider landscape, it also varied between the hillforts. Strong visual relationships are exhibited between several of the sites. For example, the Bulwark has a good degree of visibility into the interior of the West and North Enclosure on Hardings Down. There is also good visibility of the Bulwark from Hardings Down, but this is restricted to its southern enclosing works. Although fieldwork photography highlighted that visual clarity towards these enclosing works was poor, there still remains to be the potential that the morphological orientation of these enclosing works was motivated by the presence of hillforts on Hardings Down and/or the downland to the south.

Whilst the visual accessibility of the enclosing works to several hillforts was highly variable, and in some cases, there were restricted opportunities to see the true extent of the sites, there are several whereby the interior of their enclosures were visually open. For example, sites such as Prestonbury and the West and North Enclosures on Hardings Down, were positioned in such a way that meant that their interior was visible from the surrounding landscape.

Conversely there is also evidence for visual relationships to the contrary. The morphological orientation and viewshed results at Prestonbury and Cranbrook Castle indicate that these sites oriented their backs towards one another. These aspects were of limited enclosure and were the only parts of the sites that were visible from one another. A further example of sites within close proximity having limited intervisibility is the Hardings Down Group. 139

A GIS-based Analysis of Hillfort Location and Morphology

Figure 120 Results of cost surface analysis highlighting the routes where Cranbrook Castle’s enclosing works are most visible (Dartmoor National Park Historic Environment Record 2013; © Environment Agency copyright and/or database right 2015. All rights reserved)

Correct pathways

Here, intervisibility between these sites is restricted to their outer perimeter. These examples indicate that these sites were not portraying an image of strength towards one another which would be expected if there was an element of competition between them.

By definition, correct pathways inhibit direct entry into hillforts, they form morphologically defined areas of movement that may have been designed to influence the experience of those who were entering the sites (Driver 2013). This study investigates correct pathways from multiple perspectives, from morphologically defined pathways that mirror Driver’s perspective to GIS-based pathways that identify the topographically afforded pathways with visual and physically defined characteristics. Below the topographically afforded pathways are investigated from the perspective of how they were acknowledged and utilised within the morphological orientation of sites.

As discussed, the hillforts within this study have visual and morphological qualities which imply that in some cases there were conscious efforts made to influence people’s perception of them through the image that they portrayed. Image is also greatly affected by how people move around the landscape and approach these sites. The concept of correct pathways, the evidence for them and their influence on the perception of sites are discussed below.

140

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 121 Results of viewshed analysis indicating the visibility of Cranbrook Castle from Wooston Castle, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Morphologically defined correct pathways

entrances to the enclosures form a directed route into the site which was lined by entrances, but it was not formalised by a hollow way. The formation of this routeway formed a procession-like route through the site which is marked by different stages of entry into the enclosures. A similar processional route is evident within the Aberdeenshire test area at Dunnideer and Cairnmore. Both of these sites have multiple entranceways to their various enclosures that have the same orientation.

The key focus of this study’s investigation of correct pathways was upon those which were identified through GIS-based analysis. However, the morphology of these sites in themselves formed a correct pathway. In particular, entrance morphology had the potential to form a correct pathway. They formed morphologically induced correct pathways. An elaborate example of a morphologically induced correct pathway is demonstrated at Wooston Castle. Moving in and out of Wooston Castle was a physically controlled process due to the hollow way which approached the site from the south. This hollow way travelled downslope between the site’s first and third banks, there is the potential that this extended to between the third and fourth bank which is the southernmost bank (Quinnell, Griffith et al. 2013), Whilst the hollow way was physically restrictive it was also visually restrictive as there is limited visibility out of it, even today.

Slope-based correct pathways Over a third of the hillforts within this study had at least one entrance which strongly correlated with the slope-based least cost pathways. Whilst the entrances at Wooston Castle, Dunnideer, Cairnmore and Prestonbury formed morphologically induced correct pathways, slopebased cost surface analysis found that these entrances were also placed in the most physically accessible aspect of the site. At Dunnideer, cost surface analysis found that the most physically accessible route into the site is through the site’s north-western area which also includes one of the site’s entrances (Figure 170). Similarly, at Cairnmore

Although not as elaborate, the entrance formation at Prestonbury Castle is another example of a morphologically induced correct pathway. Here, the alignment of the various 141

A GIS-based Analysis of Hillfort Location and Morphology

Figure 122 Results of viewshed analysis from the radii towards Wheedlemont depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence)

142

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 123 View of Wheedlemont from Cairnmore (Jessica Murray 2013)

the majority of the slope-based pathways (136 out of 250) enter/exit very close to its entrance (Figure 171). Wooston Castle’s hollow way is the second most accessible route into the site as 108 slope-based least cost pathways that are based on the Ordnance Survey Profile DTM coincide with it. The most accessible point is from the north-east as 115 pathways intersect with it. Two of the LiDAR slope-based least cost pathways use the hollow way, the remaining two paths coincide with the north-east and north-west (Figure 172). At Prestonbury 228 out of the 244 Ordnance Survey DTM and 3 out of the 4 LiDAR based least cost pathways intersect with its eastern area, the LiDAR pathways intersect with the outer and middle entrances (Figure 173).

pathways which supports Jervise who believed that this is one of its most accessible areas (1871, 327). However, the majority of the slope-based pathways (154 out of 249) intersect with the site’s south-eastern aspect. This evidence indicates that ease of movement into Wheedlemont was not a fundamental influence in its morphological orientation. In fact, the evidence indicates quite the contrary, whereby the site was heavily enclosed in its most accessible areas and entry was inhibited here. Within the Dartmoor test area, slope-based cost surface analysis at Cranbrook Castle indicated that the southwestern entrance was placed very close to its most accessible area. Here, 143 out of the 247 Ordnance Survey slope-based pathways coincide (Figure 176). The eastern entrance interacts with 35 of the Ordnance Survey and 2 out of the 4 LiDAR slope-based pathways which indicates that this is another entrance that was placed within a relatively accessible area.

Although there is no evidence for an entrance within the inner enclosure at Tap o’Noth, the slope-based cost surface analysis found that 161 out of 247 slope-based pathways intersect with its south-eastern corner which coincides with its modern entry point (Figure 174). The accessibility of Tap o’Noth’s eastern side was also noted by Macdonald (1891, 39). This therefore may have been a natural entry point into the site prior to the closure of the inner fort.

In the Gower test area, the north-eastern entrance of Hardings Down West Enclosure is situated within the site’s most accessible area as 204 (out of 249) Ordnance Survey DTM based pathways and 2 (out of 6) LiDAR based pathways intersect with it (Figure 177). The second most accessible aspect of the site is its south-eastern corner as 21 Ordnance Survey based pathways and 3 LiDAR pathways use this area. This corner faces Hardings

On the contrary, cost surface analysis at Wheedlemont indicates that only 1 of the 249 slope-based pathways coincides with its north-eastern entrance (Figure 175). The site’s western aspect coincides with 69 slope-based 143

A GIS-based Analysis of Hillfort Location and Morphology

Figure 124 Results of viewshed analysis indicating the visibility of Wheedlemont from Cairnmore (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

On the other hand, at Hardings Down East Enclosure the entrance was not positioned in the most accessible area. the slope-based least cost pathways also fail to interact strongly with the site’s most enclosed aspect. The most accessible aspect of this site is in the south where 214 of the 218 slope-based least cost pathways intersect with this area. The second most accessible aspect according to the Ordnance Survey DTM is in the north-east as 18 of the slope-based least cost pathways use this area, which is very close to the entrance (Figure 178). Although this is

Down East enclosure and it is also one of the site’s most prominently defined aspects as there are multiple outworks. The morphological arrangement of Hardings Down West Enclosure indicates that slope was a highly influential factor during its construction. For example, whilst the entrance was positioned to enable movement by positioning it in the most accessible area, efforts were also made to counteract accessibility in un-prescribed areas by increasing the enclosing works in the second most accessible aspect of the site. 144

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 125 View of Wheedlemont from Tap o’Noth (Jessica Murray 2013)

the second most accessible area, based upon the frequency of the least cost pathways, it is such a small number that this is not significant and is unlikely to indicate that slope was a fundamental factor in the placement of the entrance.

that the morphology ignores the easy accessibility of the northern aspect of the site to inhibit accessibility into Battlesbury.

In the Warminster test area, the slope-based cost surface analysis at Cley Hill demonstrates that the easiest route into or out of the site is via the site’s north-eastern corner. This coincides with 107 out of 247 least cost pathways (Figure 179), they also travel past Little Cley Hill.

Over a third of the total number of hillforts within this study had entrances which correlated with the visible pathways. By definition, this meant that at least one of their entrances was placed where approaching traffic was visible. All of the hillforts from the Dartmoor test area had an entrance that exhibited this trait. At Cranbrook Castle all of the visible pathways correlate with the north-eastern aspect of the site and in many cases the entrance way (Figure 181). According to the Ordnance Survey profile DTM 4 out of the 8 visible pathways interact with the southern entrance and hollow way at Wooston Castle (Figure 182). However, the LiDAR based DTM implies that the north-eastern corner of Wooston Castle is its most visible entry point but they do not have a single entry point. This indicates that there is no overarching route into the site which is most visible.

Visible correct pathways

Whilst there were several sites within this study whereby their entrances had a distinct correlation with the slopebased pathways, evidence at Battlesbury indicates that there was a complete disregard for the accessibility of the site’s northern aspect. Colt Hoare argued that the site is most accessible from the north and south which is where additional ramparts were raised (1821). The high accessibility of Battlesbury’s northern aspect is supported by the results of cost surface analysis as the majority of slope-based least cost pathways enter/exit the site from the north (Figure 180). This area coincides with the Battlesbury Bowl settlement, but not the site’s north-western entrance. Two LiDAR based least cost pathways intersect with the site’s north-western entrance; one also corresponds with the south-eastern entrance. However, the sheer volume of slope-based pathways (244) that disregard the north-western entrance implies

At Prestonbury, half of the Ordnance Survey profile DTM least cost pathways interact with the eastern entrance (Figure 183). This is further supported by the LiDAR pathways as two interact with this entrance. Similarly, at Dunnideer half of the visible pathways intersect with the site’s entrances (Figure 184) These results further 145

A GIS-based Analysis of Hillfort Location and Morphology

Figure 126 Results of viewshed analysis indicating the visibility of Wheedlemont from Tap o’Noth (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

pathways with key architectural components exhibited at Battlesbury. Here, the pathways tend to use the southern, south-eastern and south-western aspect of the site (Figure 186). This sector is one of its most defended aspects, which is not the most accessible. A similar pattern was also exhibited at the Bulwark where all of the pathways intersect with the site’s southern enclosing works which is its most enclosed area (Figure 187), The highly enclosed nature of these areas may have been placed to face an area where people were known to approach from.

imply that one of the key motivations behind placing the entrance to this site was that they knew that approaching traffic were highly visible from this area. In the Warminster test area, cost surface analysis at Scratchbury indicates that the north-western entrance is situated where approaching traffic is most visible. Here, 4 out of the 11 Ordnance Survey Profile DTM pathways and 1 of the 4 LiDAR pathways coincide with the area (Figure 185). There is a much weaker pattern of the visible 146

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 127 Results of viewshed analysis indicating the visibility of Wheedlemont from Dunnideer (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Blind correct pathways

at Battlesbury, there is a high frequency of the blind pathways using the south-eastern entrance; here 7 out of the 12 Ordnance Survey PROFILE DTM and 3 out of the 4 LiDAR pathways coincide with it (Figure 189).

There are a number of examples whereby blind pathways correlate with key architectural components at the hillforts within this study. The results from cost surface analysis at two of these sites indicate that at least one of their entrances was placed where the site was least visible upon approach. The first example is Hardings Down North Enclosure. Here, the entrance corresponds with a third of the LiDAR blind pathways (Figure 188). Similarly,

Interestingly, at Hardings Down West Enclosure the most enclosed aspect of the site is also its least visible upon approach (Figure 190). Here 7 out of 18 Ordnance Survey based DTM pathways and 3 out of 6 LiDAR pathways coincide with the north-eastern aspect. Similarly, at 147

A GIS-based Analysis of Hillfort Location and Morphology

Figure 128 View of Wheedlemont from the surrounding landscape. Map A depicts the view from Viewpoint 17, Map B depicts the view from Viewpoint 12 , Map C depicts the view from Viewpoint 11, whilst Map D depicts the location of the Viewpoints (Aberdeenshire Council Archaeology Service 2014; Jessica Murray 2013; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence)

148

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 129 Viewshed results indicating the visibility of Cairnmore from the surrounding radii (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

149

A GIS-based Analysis of Hillfort Location and Morphology

Figure 130 Results of viewshed analysis indicating the visibility of Cairnmore from Wheedlemont (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Cairnmore 9 out of the 18 blind pathways coincide with the most enclosed aspect of the site in the west (Figure 191). These examples echo the evidence found at a number of the Ceredigion hillforts which was detailed in Chapter 3.

this hinders the visibility to the site. It is therefore unlikely to indicate anything of archaeological significance. The strongest pattern of blind pathways exhibited within this study was at Wheedlemont whereby all of the 18 pathways coincide with the site’s south-western aspect (Figure 193). However, as there is no strong evidence for morphological variation at this site, there is no indication that the site portrayed a particular image in this direction.

At Cley Hill the blind pathways enter and exit the site primarily via its south-western aspect (Figure 192). This is likely influenced by the quarrying in the south-west, as 150

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 131 View of Cairnmore from Wheedlemont (Jessica Murray 2014)

Another significant route that corresponds with the visible pathways are those between Dunnideer and Cairnmore (Figure 195). The concentration of both later prehistoric and early medieval activity (II) includes the enclosure at Barflat (NJ42NE0047). Excavations revealed that this comprised concentric enclosures which were centred on a ring ditch which was in association with the Craw Stane (Noble and Gondek 2011). Finds of late Roman Amphorae, 6th Century continental glass, metalwork and metalworking debris imply that this was a high status 6th Century Settlement (Ibid. 2011). Subsequent excavations found multiple objects associated with metalworking that further supported the suggestion that this site was of high status and was possibly a Royal site that served both a ritual and settlement function (Noble et al. 2012). These pathways also come across square barrows (NJ42NE0057) and two further enclosures (NJ42NE0060). The most significant correlation of the visible pathways with known routeways travelled between Dunnideer and the surrounding landscape was found at III. This routeway travels via Wardhouse Hill. Although this area is not populated by archaeological activity, there is an extant trackway (NJ53SE0054) of unknown date. This track may have been a significant routeway in the past, a significant route that was also visible from Dunnideer.

This does however contradict Jervise’s observation that the north-eastern entrance faced a blind hillside (1871). The wider archaeological context Although cost surface analysis identified various significant trends which indicate that slope and visual qualities influenced the morphological orientation of various hillforts, in a wider context it identified potentially significant routeways through the landscape. An assessment of the cost surface analysis found a correlation between the least cost pathways and known archaeological sites, which were likely contemporary or at least extant at the same time that the hillforts were in use, in the Dartmoor and Aberdeenshire test areas. There is strong evidence for the slope and visible pathways within the Aberdeenshire test area coinciding with areas of known archaeological activity. For example, the visible pathways to and from Wheedlemont (Figure 194) travel close to a boulder with at least 10 cupmarks (NJ42NE0043), a possible late Bronze Age hoard (NJ42NE0032) and copper coins (NJ42NE0013) (I). Even though it was implied by Cook that Wheedlemont dated to the early Medieval period (2013), the activity around this site was dense within the later prehistoric period. This implies that within later prehistory this was a significant place, which may have factored into communication routes within this landscape.

The visible pathways which travel between Tap o’Noth and the neighbouring hillforts do not distinctively follow the line of topographic features. However, they do interact with 151

A GIS-based Analysis of Hillfort Location and Morphology

Figure 132 Results of viewshed analysis indicating the visibility of Cairnmore from Tap o’Noth (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

a small number of sites. The visible pathways between Tap o’Noth and Cairnmore travel close to and through Rhynie, which has a dense concentration of both later prehistoric and early medieval activity (I) (Figure 196). This activity includes the enclosure at Barflat (NJ42NE0047), square barrows (NJ42NE0057), two further enclosures (NJ42NE0060), two standing stones (NJ42NE0022), two Pictish symbol stones (NJ42NE0021) and two urns (NJ42NE0020).

test area, there was also a strong correlation of the slopebased pathways (Figure 197). At Dunnideer, the slope paths to Wheedlemont and Tap O’Noth largely follow the same line and initially follow the course of the River Shevock and the railway line to Kennethmont. Beyond this, the pathways follow the line of several waterways where they come across a series of enclosures, ring ditches and standing stones (I).

Whilst there was a strong correlation of visible pathways with known archaeological sites within the Aberdeenshire

At Wheedlemont, the path from Tap o’Noth interacts with a number of archaeological features such as a boulder with 152

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 133 Results of viewshed analysis indicating the visibility of Cairnmore from Dunnideer (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

The paths between Wheedlemont and Dunnideer interact with several archaeological features which are close to Wheedlemont (III) (Figure 198). These features include 2 enclosures (NJ42NE0060), a D-shaped enclosure (NJ42NE0056), a cluster of prehistoric finds and two standing stones (NJ52NW0003). These pathways subsequently follow the line of numerous waterways which include the River Shevock. The pathways also follow the line of the modern railway line, which runs

at least 10 cupmarks (NJ42NE0043), a possible late Bronze Age hoard (NJ42NE0032) and copper coins (NJ42NE0013) (I) (Figure 198). The pathways between Cairnmore and Wheedlemont travel close to a D-shaped cropmark (NJ42NE0056) and two enclosures (NJ42NE0060). They also follow the line of the Water of Bogie and interact with several symbol stones (NJ42NE0029, NJ42NE0045 and NJ42NE0033) and Barflat enclosure (NJ42NE0047) (II). 153

A GIS-based Analysis of Hillfort Location and Morphology

Figure 134 Results of viewshed analysis indicating the visibility of Dunnideer from the surrounding radii (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

154

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 135 View of Dunnideer from the surrounding landscape. Map A depicts the view from Viewpoint 31, Map D depicts the view from Viewpoint 27, Map C depicts the view from Viewpoint 26, whilst Map D illustrates the location of the Viewpoints (Aberdeenshire Council Archaeology Service 2014; Jessica Murray 2013; Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

155

A GIS-based Analysis of Hillfort Location and Morphology

Figure 136 Viewshed results indicating the visibility of Dunnideer from Tap o’Noth (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

close to Dunnideer, the establishment of the railway within this area indicates that it is an important route through this landscape even today.

number of ring ditches (NJ52NW0029 and NJ52NW0030) and a rectilinear enclosure (NJ52NW0041) (II). The pathways continue their journey by roughly following and running parallel with several watercourses and the modern railway line towards Dunnideer.

At Tap o’ Noth the pathways between itself and Cairnmore and Dunnideer follow the line of Glamlach Burn into the lowlands (Figure 199). From this point the pathways take a different route. The paths between Tap o’ Noth and Dunnideer run parallel to the Water of Bogie. Whilst following this line the pathways travel within 120m of a

To the south of Rhynie the pathways between Tap o’Noth and Cairnmore follow the line of the Water of Bogie (III). Whilst following this, they coincide with several archaeological sites which are primarily symbol stones 156

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 137 Results of viewshed analysis indicating the visibility of Dunnideer from Wheedlemont (Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

(NJ42NE0029, NJ42NE0045 and NJ42NE0033) and enclosures (NJ42NE0047). The dense concentration of Pictish symbol stones within the Rhynie area implies that this was a highly significant place within the Early Medieval period (Fraser and Halliday 2008, 119). Fraser and Halliday highlight that Rhynie was situated on an influential north-south routeway through the hills, which provided access to the Moray coast (2008, 119-120). The pathway from Tap o’Noth towards Wheedlemont interacts with a small number of known sites these include a boulder

with at least 10 cupmarks (NJ42NE0043), a possible late Bronze Age hoard (NJ42NE0032) and copper coins (NJ42NE0013) (I). At Cairnmore, the initial routes of the slope paths between itself and Dunnideer do not distinctively follow any topographically defined routes (Figure 200). At various stages they briefly follow the line of a watercourse. To the east of Clatt, the pathway travels through a large circular enclosure (A). The pathways 157

A GIS-based Analysis of Hillfort Location and Morphology

Figure 138 Results of viewshed analysis indicating the visibility of Dunnideer from Cairnmore (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

towards Wheedlemont and Tap o’Noth intersect with both topographical and archaeological features. They follow the line of the Water of Bogie and come into contact with several enclosures (NJ42NE0047) and symbol stones (NJ42NE0045, NJ42NE0033 and NJ42NE0029) (III). At Rhynie the pathways take different routes. The pathways between Wheedlemont and Cairnmore follow the line of several water ways and enclosures (NJ42NE0060 and NJ42NE0056) (I). The paths between Cairnmore and Tap o’ Noth do not travel past any further archaeological

features but they do follow the line of Glamlach Burn (II). This implies that this area may not have been a significant route in the past or more recent activity has removed all traces of activity. In Dartmoor, there are a number of instances whereby the visible pathways correlated with known activity within the Dartmoor study area. The visible pathways between Cranbrook and Wooston Castle coincide with activity which is likely to have been contemporary with the hillforts 158

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 139 Results of viewshed analysis indicating the visibility of Tap o’Noth from the surrounding radii when using the Ordnance Survey PROFILE DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

as they travel past two ovoid enclosures (MDV26738 and MDV26739), a bank (MDV26742) (I) and Prestonbury Castle (Figure 201-202). This routeway is therefore highly visible from both Cranbrook and Wooston Castle, but they were also highly significant places within this landscape due to the substantial investment in them.

The morphological orientation of hillforts and their entranceways formed morphologically defined correct pathways to varying degrees. For example, at Wooston Castle the hollow way formed a well-defined route into the site that did not promote movement beyond it. Further evidence for morphologically defined correct pathways is found at the multiple enclosure sites such as Prestonbury, Cairnmore, Dunnideer and Tap o’Noth whereby the entrances to their different enclosures were positioned adjacent to one and other, however movement beyond this line may still have been possible.

Summary Movement was highly influential within the construction and morphological orientation of hillforts. However, as with the topographical relationship of the morphologies, their visual qualities and the images that they portrayed, the investigation into correct pathways also illustrated a very variable picture.

In the case of Battlesbury, an incorrect pathway was evident. Here, the north-western entrance was positioned offset from the plateau to the north which was the natural 159

A GIS-based Analysis of Hillfort Location and Morphology

Figure 140 View of Tap o’Noth from the surrounding landscape. Map A depicts the view from Viewpoint 3, Map B depicts the view from Viewpoint 10, Map C depicts the view from Viewpoint 12 whilst Map D depicts the location of the Viewpoints (Aberdeenshire Council Archaeology Service 2014; Jessica Murray 2013; Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

160

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 141 Results of viewshed analysis indicating the visibility of Tap o’Noth from Dunnideer when using the Ordnance Survey PROFILE DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

also naturally induced correct pathways which enabled and promoted movement. Additionally, over a third of the hillforts had entrances that correlated with the visible pathways, which demonstrates that they were positioned in such a way that they had maximum visual connectivity with the approaching traffic. In this respect the entrance placement and morphology was oriented towards a known audience.

landward approach. This increased the complexity of entering Battlesbury which indicates that the morphological orientation of this hillfort was not designed with functionality in mind. Only a third of the test hillforts had an entrance which was placed at the easiest entry point into the site. These entranceways were often elaborate and went beyond a simple gap in the enclosure circuit. This demonstrates that morphologically enhanced correct pathways were

Chapter 3 indicated that several hillforts within the 161

A GIS-based Analysis of Hillfort Location and Morphology

Figure 142 View of Tap o’Noth from Dunnideer (Jessica Murray 2013)

Ceredigion test area had their most prominently enclosed aspect facing an area which correlated with the blind pathways. This was not reflected on such a large scale within the remaining test areas, however it was reflected at Wheedlemont and Hardings Down West Enclosure. Although it was not as strong an occurrence as it was in the Ceredigion test area, it does however further highlight the complexity of the motivations behind the morphological orientation of hillforts, which has been highlighted throughout this chapter and will be summarised below.

the only factor. In these instances, topographical prominence caused the hillforts to be visually and physically prominent whereas the enclosing works did not. At Cley Hill and Scratchbury, whilst the sites were large, the enclosing works were not visually prominent. Similarly, in the Aberdeenshire test area the outer forts at Tap o’Noth and Dunnideer, and the enclosing works at Cairnmore and Wheedlemont have limited physical prominence but the topographical landforms that they were positioned on do. There are however several other examples whereby the physical and visual prominence of sites is enhanced by the physical contrast of the enclosing works with the underlying topography. This was evident at the inner forts of Tap o’Noth and Dunnideer. Both of their sites are oblong in shape. The morphology of these inner enclosures was not dictated by the rounded, irregular shape of the hilltops, they were designs that were enforced upon the landscape.

Conclusion The topographical nature of the test areas was variable. In general, two are highly topographically defined, whilst the remaining two are less topographically prominent. The Dartmoor test area is a topographically divided landscape that is focused on the Teign river valley. Another topographically prominent region is the Aberdeenshire test area. This comprises low lying river valleys and large hill range’s such as the Coreen Hills, Wardhouse Hill and the Hill of Noth. Whilst the Gower is a less topographically prominent region, it is topographically diverse and comprises a combination of coastal plains, ridges and isolated hills. The Warminster test area is a rolling downland that borders the edge of Salisbury Plain

Further evidence for the conscious design and orientation of some of the hillforts within this study is demonstrated at the Bulwark, Hardings Down West Enclosure and Battlesbury. These sites had disproportionate vallation sequences that faced potentially contemporary sites which implies that their morphologies were directed towards known and significant audiences, potentially to influence people’s perception of them. Conversely, in the Dartmoor test area, Cranbrook and Prestonbury Castle oriented their least enclosed elements towards one another which

Topography is a key influencing factor in the physical and visual prominence of hillforts, and at several it is seemingly 162

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 143 Results of viewshed analysis indicating the visibility of Tap o’Noth from Cairnmore when using the Ordnance Survey PROFILE DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

163

A GIS-based Analysis of Hillfort Location and Morphology

Figure 144 Results of viewshed analysis indicating the visibility of Tap o’Noth from Wheedlemont when using the Ordnance Survey PROFILE DTM (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

164

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 145 View of Tap o’Noth from Wheedlemont (Jessica Murray 2013)

Figure 146 View of Tap o’Noth from Cairnmore (Jessica Murray 2013)

165

A GIS-based Analysis of Hillfort Location and Morphology

Figure 147 Results of viewshed analysis from the radii towards Hardings Down North Enclosure depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (DTM for analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

166

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 148 View of Hardings Down North Enclosure from the surrounding landscape. Map A depicts the view from Viewpoint 2, Map B depicts the view from Viewpoint 3, whilst Map C illustrates the location of the Viewpoints (Information held by the GGAT HER Charitable Trust copyright (provided 2014); Jessica Murray 2013; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

was limited potential for them to portray a complete image towards the surrounding landscape. This meant that access to visual information within the landscapes that surrounds them was limited. Consequently, whilst there was large scale morphological investment in them, their visual impact was not proportionate to this investment.

implies that here there was no desire to impress or deter those within the neighbouring sites. The evidence for variable degrees of physical and visual prominence is further echoed at Prestonbury, Cranbrook and Hardings Down North Enclosure where there was a distinct lack of visual and physical prominence. The morphology of these sites indicates that their enclosing works acted to solely enclose space as opposed to create a powerful and imposing structure.

Just as the enclosing works at the test hillforts exhibited varying degrees of visual and physical qualities, as too did their interiors. Several sites had visually open interiors. The interior of the hillslope enclosures of Prestonbury, Hardings Down North Enclosure and Hardings Down West Enclosure, were highly visible. Similarly, although Scratchbury and Cley Hill are not hillslope enclosures, siting the enclosing works downslope from the main enclosure meant that their interiors were visible.

Whilst varying degrees of physical and visual prominence was exhibited across the study areas, the visual qualities of the enclosing works at many sites also exhibited intra-site variation. Cost surface analysis indicated that the majority of the sites portrayed very fragmentary images as there 167

A GIS-based Analysis of Hillfort Location and Morphology

Figure 149 Results of viewshed analysis indicating the visibility of Hardings Down North Enclosure from Hardings Down West Enclosure, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

deliberate disregard for enabling easy access into the site, over a third of the remaining hillforts within the study had at least one entrance that was placed in one of their most accessible aspects.

Whilst the visual qualities of hillforts varied, as too did the means by which people could move in and out of them. Whilst this study focused upon investigating the correlation of GIS-based least cost pathways with known morphological and topographical elements, there were also several instances of morphologically defined correct pathways. Most prominently this was seen at Wooston Castle with the hollow way which also correlated with one of the most accessible aspects of the site. Similarly, at Prestonbury, Cairnmore and Dunnideer multiple entrances were aligned towards one another which formed a processional route into the sites.

In terms of the visual pathways, over a third of the hillforts had entrances that correlated with the majority of them. This indicates, in these cases, that entrances were placed where the approaching traffic was most visible. Conversely, several hillforts had entrances that correlated with the blind pathways. This demonstrates that they were placed where the site was least visible from approaching traffic. Additionally, just as with a number of the Ceredigion test hillforts, at Hardings Down West Enclosure and Wheedlemont their most enclosed aspects correlated with the blind pathways. The orientation of morphological investment such as entrances and disproportionate vallation sequences in the least visible aspect of the hillfort upon approach enhanced first impressions of these sites. Consequently, it greatly influenced how people, who were

Although there was evidence for the elaboration of movement in and out of sites, at Battlesbury there was evidence for the complete disregard for ease of access. Here the north-western entrance was positioned offset from the site’s most accessible point. This area also coincided with the Battlesbury Bowl settlement which is thought to be the predecessor to the hillfort. Whilst Battlesbury exhibited a 168

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 150 Results of viewshed analysis indicating the visibility of Hardings Down North Enclosure from Hardings Down East Enclosure, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

visiting these sites for the first time, perceived them and the community that was within them.

meant that movement and occupation within these areas was influenced by them and was likely more concentrated. Whereas, in the more subtle landscapes of the Gower and Warminster test areas, people were less likely to have been drawn to locations based upon their topographical features alone. In these instances, activity is more likely to have been sporadic and more widely spaced.

Whilst many of the GIS-based least cost pathways correlated with key aspects of hillfort morphologies, in the case of the Aberdeenshire and Dartmoor test areas many of the least cost pathways also correlate with known areas of archaeological activity. This indicates that the GIS-based least cost pathways may in fact coincide with established routeways that were in existence at the same time that the hillforts were in use. Both the Aberdeenshire and Dartmoor test areas are topographically prominent regions with highly distinct topographical features. The prominence of these topographical features inevitably

To conclude, this chapter has highlighted a high degree of variability in the visual and physical qualities of the test hillforts. These qualities will be explored further in Chapter 5, where the findings will be discussed within the wider context of the potential use of hillforts.

169

A GIS-based Analysis of Hillfort Location and Morphology

Figure 151 Results of viewshed analysis indicating the visibility of Hardings Down North Enclosure from The Bulwark, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Figure 152 View of Hardings Down North Enclosure from The Bulwark (Jessica Murray 2013)

170

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 153 Results of viewshed analysis from the radii towards Hardings Down East Enclosure depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (© Crown Copyright and Database Right (2018). Ordnance Survey (DTM for analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

171

A GIS-based Analysis of Hillfort Location and Morphology

Figure 154 Results of viewshed analysis indicating the visibility of Hardings Down East Enclosure from Hardings Down North Enclosure, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Figure 155 Results of viewshed analysis indicating the visibility of Hardings Down East Enclosure from Hardings Down West Enclosure, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

172

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 156 Results of viewshed analysis indicating the visibility of Hardings Down East Enclosure from The Bulwark, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence)

Figure 157 View of Hardings Down East Enclosure from The Bulwark (Jessica Murray2013)

173

A GIS-based Analysis of Hillfort Location and Morphology

Figure 158 Results of cost surface analysis highlighting the routes where Hardings Down East Enclosure’s enclosing works are most visible (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; information held by the GGAT HER Charitable Trust copyright (provided 2014))

174

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 159 Results of viewshed analysis from the radii towards Hardings Down West Enclosure depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (DTM for analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved)

175

A GIS-based Analysis of Hillfort Location and Morphology

Figure 160 View of Hardings Down West Enclosure from the surrounding landscape. Map A depicts the view from Viewpoint 2, Map B depicts the view from Viewpoint 3, Map C depicts the view from Viewpoint 1, whilst Map D illustrates the location of the Viewpoints (Information held by the GGAT HER Charitable Trust copyright (provided 2014); Jessica Murray 2013; Basemapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

176

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 161 Results of viewshed analysis indicating the visibility of Hardings Down West Enclosure from Hardings Down East Enclosure, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Figure 162 View of Hardings Down West Enclosure from Hardings Down East Enclosure (Jessica Murray2013)

177

A GIS-based Analysis of Hillfort Location and Morphology

Figure 163 Results of viewshed analysis indicating the visibility of Hardings Down West Enclosure from Hardings Down North Enclosure, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Figure 164 Results of viewshed analysis indicating the visibility of Hardings Down West Enclosure from The Bulwark, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

178

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 165 Results of cost surface analysis highlighting the routes where Hardings Down West Enclosure’s enclosing works are most visible (Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; information held by the GGAT HER Charitable Trust copyright (provided 2014))

179

A GIS-based Analysis of Hillfort Location and Morphology

Figure 166 Results of viewshed analysis from the radii towards Wooston Castle depicting the visibility of the site when using the Ordnance Survey PROFILE DTM (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

180

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 167 Results of viewshed analysis indicating the visibility of Wooston Castle from Cranbrook Castle, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Figure 168 Results of viewshed analysis indicating the visibility of Wooston Castle from Prestonbury Castle, Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst MAP B depicts the results when using the 1m resolution LiDAR (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

181

A GIS-based Analysis of Hillfort Location and Morphology

Figure 169 Results of cost surface analysis highlighting the routes where Wooston Castle’s enclosing works are most visible (Dartmoor National Park Historic Environment Record 2013; © Environment Agency copyright and/or database right 2015. All rights reserved)

182

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 170 Results of slope-based cost surface analysis indicating the accessibility of Dunnideer (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

183

A GIS-based Analysis of Hillfort Location and Morphology

Figure 171 Results of slope-based cost surface analysis indicating the accessibility of Cairnmore (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

184

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 172 Results of slope-based cost surface analysis indicating the accessibility of Wooston Castle (© Environment Agency copyright and/or database right 2015. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Figure 173 Results of slope-based cost surface analysis indicating the accessibility of Prestonbury Castle (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

185

A GIS-based Analysis of Hillfort Location and Morphology

Figure 174 Results of slope-based cost surface analysis indicating the accessibility of Tap o’Noth (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

186

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 175 Results of slope-based cost surface analysis indicating the accessibility of Wheedlemont (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

187

A GIS-based Analysis of Hillfort Location and Morphology

Figure 176 Results of slope-based cost surface analysis indicating the accessibility of Cranbrook Castle. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 1m resolution LiDAR DTM (Dartmoor National Park Historic Environment Record 2013; © Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

Figure 177 Results of slope-based cost surface analysis indicating the accessibility of Hardings Down West Enclosure. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 2m resolution LiDAR DTM (Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

188

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 178 Results of slope-based cost surface analysis indicating the accessibility of Hardings Down East Enclosure. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 2m resolution LiDAR DTM (Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

189

A GIS-based Analysis of Hillfort Location and Morphology

Figure 179 Results of slope-based cost surface analysis indicating the accessibility of Cley Hill. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 1m resolution LiDAR DTM (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)

190

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 180 Results of slope-based cost surface analysis indicating the accessibility of Battlesbury. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 1m resolution LiDAR DTM (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)

191

A GIS-based Analysis of Hillfort Location and Morphology

Figure 181 Sum of visible least cost pathways at Cranbrook Castle. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 1m resolution LiDAR DTM (Dartmoor National Park Historic Environment Record 2013; © Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

192

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 182 Sum of visible least cost pathways at Wooston Castle. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 1m resolution LiDAR DTM (Dartmoor National Park Historic Environment Record 2013; © Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

193

A GIS-based Analysis of Hillfort Location and Morphology

Figure 183 Sum of visible least cost pathways at Prestonbury Castle. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 1m resolution LiDAR DTM (Dartmoor National Park Historic Environment Record 2013; © Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

194

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 184 Sum of visible least cost pathways at Dunnideer (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

195

A GIS-based Analysis of Hillfort Location and Morphology

Figure 185 Sum of visible least cost pathways at Scratchbury. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 1m resolution LiDAR DTM (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence) ; Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)

196

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 186 Sum of visible least cost pathways at Battlesbury. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 1m resolution LiDAR DTM (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence) ; Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)

197

A GIS-based Analysis of Hillfort Location and Morphology

Figure 187 Sum of visible least cost pathways at the Bulwark. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 2m resolution LiDAR DTM (Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

198

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 188 Sum of blind least cost pathways at Hardings Down North Enclosure. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 2m resolution LiDAR DTM (Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

199

A GIS-based Analysis of Hillfort Location and Morphology

Figure 189 Sum of blind least cost pathways at Battlesbury. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 1m resolution LiDAR DTM (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence); Archaeology data derived from the Wiltshire and Swindon Historic Environment Record © Wiltshire Council 2018)

200

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 190 Sum of blind least cost pathways at Hardings Down West Enclosure. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 2m resolution LiDAR DTM (Derived from information held by the GGAT HER Charitable Trust copyright (provided 2014); Contains Natural Resources Wales information © Natural Resources Wales and database right. All rights reserved; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

201

A GIS-based Analysis of Hillfort Location and Morphology

Figure 191 Sum of blind least cost pathways at Cairnmore (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

202

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 192 Sum of blind least cost pathways at Cley Hill. Map A depicts the results when using the Ordnance Survey PROFILE DTM whilst Map B illustrates the results from the 1m resolution LiDAR DTM (© Environment Agency copyright and/or database right 2015. All rights reserved.; DTM for Map A analysis © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence)

203

A GIS-based Analysis of Hillfort Location and Morphology

Figure 193 Sum of blind least cost pathways at Wheedlemont (Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

204

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 194 Landscape wide distribution of visible least cost pathways to and from Wheedlemont. Map A depicts the general distribution whilst Map B is a zoom in on the correlation of the least cost pathways with known archaeology (Aberdeenshire Council Archaeology Service 2014; Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

205

A GIS-based Analysis of Hillfort Location and Morphology

Figure 195 Landscape wide distribution of visible least cost pathways to and from Dunnideer and their correlation with known archaeology (Aberdeenshire Council Archaeology Service 2014; Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

206

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 196 Landscape wide distribution of visible least cost pathways to and from Tap o’Noth. Map A depicts the general distribution whilst Map B is a zoom in on the correlation of the least cost pathways with known archaeology (Aberdeenshire Council Archaeology Service 2014; Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

207

A GIS-based Analysis of Hillfort Location and Morphology

Figure 197 Landscape wide distribution of slope-based least cost pathways to and from Dunnideer. Map A depicts the general distribution whilst Map B is a zoom in on the correlation of the least cost pathways with known archaeology (Aberdeenshire Council Archaeology Service 2014; Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

208

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 198 Landscape wide distribution of slope-based least cost pathways to and from Wheedlemont. Map A depicts the general distribution whilst Map B is a zoom in on the correlation of the least cost pathways with known archaeology (Aberdeenshire Council Archaeology Service 2014; Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

209

A GIS-based Analysis of Hillfort Location and Morphology

Figure 199 Landscape wide distribution of slope-based least cost pathways to and from Tap o’Noth. Map A depicts the general distribution whilst Map B is a zoom in on the correlation of the least cost pathways with known archaeology (Aberdeenshire Council Archaeology Service 2014; Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

210

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 200 Landscape wide distribution of slope-based least cost pathways to and from Cairnmore. Map A depicts the general distribution whilst Map B is a zoom in on the correlation of the least cost pathways with known archaeology (Aberdeenshire Council Archaeology Service 2014; Base mapping © Crown Copyright and Database Right (2018). Ordnance Survey (Digimap Licence))

211

A GIS-based Analysis of Hillfort Location and Morphology

Figure 201 Landscape wide distribution of visible least cost pathways to and from Wooston Castle and its correlation with known archaeology (Dartmoor National Park Historic Environment Record 2013; © Environment Agency copyright and/or database right 2015. All rights reserved.)

212

Multi-regional GIS-based Analysis of Hillfort Location and Morphology

Figure 202 Landscape wide distribution of visible least cost pathways to and from Cranbrook Castle and its correlation with known archaeology (Dartmoor National Park Historic Environment Record 2013; © Environment Agency copyright and/or database right 2015. All rights reserved.)

213

5 Conclusion Introduction

question. For example, it examined whether topography was used to aid construction or whether it disregarded the topographical affordances.

This study arose from the continued dissatisfaction with the term hillfort (Bedwin 1984; Harding 2012). Scholars increasingly acknowledged that there was much variation in site locations and morphologies, consequently a blanket term like ‘hillfort’ is more of a ‘hindrance than a help’ (Bedwin 1984). There has also been a persistent discontentment with the typologies which have been applied to this class of monument. Scholars recognise that there is intra-site morphological variation (Rivet 1961), and this was examined by Driver in Ceredigion (2013) and his approach was a principal influence on this study.

• Image According to Driver, image was portrayed by disproportionately allocating resources within a particular direction. This book investigated the degree to which the concept of ‘image’ was visible through Viewshed and Cost Surface Analysis. • Correct pathways

Whilst this work adopted Driver’s principle of investigating morphological variation at individual sites, it also developed the current trend for studies to apply GISbased analysis to archaeological investigations. This enabled an assessment of large quantities of data in a relatively objective manner. This study’s focus of GISbased analysis was on the investigation of movement and visibility. By moving away from earlier militaristic studies of hillforts (Dyer 1992; Sutton 1966; Hogg 1979; Fox 1961; Westropp 1896/1901; Crawford & Keiller 1928), it analysed hillforts as physical entities, whose construction was potentially motivated by more fundamental factors such as movement, visibility and topography.

Driver defined correct pathways into and around sites on a morphological basis. This book investigated GIS-based correct pathways and their correlation with morphological examples. More specifically the aim of this study was to apply a combination of both viewshed and cost surface analysis to investigate the following questions: • What was the degree of visibility of the surrounding landscape from the hillfort from different points on/ within it? Were there long distance or short distance views? • How visible were the interiors of the hillforts from other hillforts? Was one area of the hillfort more visible than others? How does the variable internal visibility relate to the morphology of the hillforts? • How did the orientation of least cost paths relate to the morphology of the banks, ditches and entrances of the hillforts? • How did movement along such paths affect the visibility of the hillforts?

The theoretical underpinnings for this study’s focus upon movement and visibility were primarily based on Gibson’s theory that an individual’s visual understanding is due to their role as a perceiver within a particular environment (1979). However, for people to be able to grasp a visual understanding of an environment there needs to be a ‘will to be visible’ (Gibson 1979; Wheatley and Gillings 2000). In the case of hillfort builders, this manifests itself through a number of ‘strategies of visibility’ (Wheatley and Gillings 2000). Movement within these landscapes meant that individuals engaged with and learnt from it (Ingold 2000; Ingold 2011). Through wayfaring, people continuously acquired knowledge just as movement was a continuous process (Ingold 2011).

Physical relationship of hillfort morphologies and locations with the landscape topography Some hillforts occupy prominent landmarks perhaps with the express intention of making a statement through high visibility; others seem to be designed with the opposite objective of merging into the natural landscape (Harding 2012, 15).

Although this study’s methodology is primarily a quantitative analysis, it was used to answer and investigate the qualitative elements that were raised by Driver’s investigation (2005a; 2013):

Driver examined how the morphology of hillforts in Ceredigion related to topography; he investigated whether their morphologies conformed with it or whether their design was enforced upon the landscape (2005a; 2007; 2013). At Castell Grogwynion, he noted that to construct the northern façade the underlying bedrock had to be quarried which would have been a very labour intensive

• The physical relationship of hillfort morphologies and locations with the landscape topography This investigation involved examining how topography influenced the location and morphology of the hillfort in 214

Conclusion process (2005a; 2013). This demonstrates that during the construction of hillforts, the least costly means were not always followed (Driver 2007). Driver also argued that at sites where there were instances of the rampart lines failing to conform with the topography, they symbolised a higher level of power and exclusivity (Ibid.). However, in the case of Castell Grogwynion, the expensive undertaking of the northern façade was justified. GIS-based analysis indicated that this façade was placed within the site’s most physically accessible area (Figure 38). This is one of eight hillforts within this study whose most extensive enclosing works were oriented towards their most accessible area. These enclosing works were placed to counteract the site’s susceptibility to easy access from outsiders, which resulted from the gentle slopes of the topography.

Directing morphological investment towards known audiences is also demonstrated at several sites where their most extensive enclosing works coincide with the visible pathways. At the Bulwark and Battlesbury, even though the visible paths do not intersect with the site’s entrance, they coincide with some of their most extensive enclosing works (Figures 185-186). These results demonstrate that some of the most extensive and elaborate morphological aspects to these hillforts were one of their most visible points. This indicates that the morphological orientation of some hillforts may have been designed to display towards a known audience equally it may have been a response to a topographical weakness. The nature of responses to topographical weaknesses is discussed below, particularly in relation to both adequate and ostentatious responses.

Similarly, at Gaer Fawr, the site’s northern side had an additional bank which faced the land that the site sat level with. This additional bank formed an unbroken façade that faced one of its most accessible aspects as 103 slope paths interact with it (Figure 42). Viewshed analysis indicated that the construction of this bank did not increase the site’s visibility, as it was no more visible than any other aspect (Figure 34-35). Consequently, the addition of a further bank appears to have had defensive motivations as opposed to ones of display. Only three out of the twentytwo test hillforts have evidence that implies that resources were disproportionately applied to their most visible aspect.

Ostentatious and adequate responses to landscape qualities Above it was highlighted that there was strong evidence for disproportionate vallation sequences at a number of hillforts within this study. The motivations behind their placement varied, for example some were placed within their most visible area and others in their most physically accessible area. However, the scale of these disproportionate vallation sequences varied; some were ostentatious whilst others adequately responded to the topography. Eleven out of the twenty-two test hillforts had morphologies which were examples of an adequate response to topography. whilst the remainder were ostentatious. The latter could represent examples of the portrayal of a particular image within a particular direction. The degree to which sites responded did not have a direct correlation with the form of its topographical environs. For example, in the most prominent landscape within this study; the deeply incised Teign River Valley, or the smaller micro landscapes which surround the sites in Ceredigion, the natural prominence of the landscape was used to aid hillfort construction. For example, at Prestonbury the site’s eastern enclosing works, which sit adjacent to the adjoining Broadmoor Common, lack visual prominence and formed a simple physical spatial delineator. The remainder of this site’s enclosing works were limited, and followed the line of the steep valley edge. On the other hand, Trecoll’s northern façade faced the only direction from which people could approach the site, and this façade was highly visually and physically prominent. Just as with Prestonbury, the remainder of the site’s enclosing works were limited, and it used the natural prominence of the topography. Trecoll’s northern façade formed an ostentatious display piece that also physically separated the site from the surrounding landscape, whereas the banks at Prestonbury were an adequate response to a need to define space.

The construction of the additional enclosing works at these sites was an expensive undertaking that were directed towards known areas of approaching traffic. Whilst they were no more visible from the wider landscape than any other aspect, they were placed to inhibit entry into the site and to be exposed to approaching traffic. These instances of disproportionate vallation sequences imply that whilst there was an emphasis on defining space within later prehistoric society, it went beyond forming a simple enclosure. The large-scale investment into these disproportionate vallation sequences, towards likely areas of movement, indicates that there was a strong desire to influence outsider’s experiences and perceptions of these sites. Placing the most impressive aspect of the site within an area where there is likely to be a large audience indicates that this investment was successful in forming a platform to influence outsider’s perceptions. The desire to influence people’s perceptions of these sites, particularly towards known audiences, was also emphasised by the placement of elaborate entrances within an area where there was maximum visibility to those that approached. The placement of these morphologically elaborate entrances where the approaching traffic was most visible indicates that those within the hillfort were visually aware of what was going on within the surrounding landscape, and they wished to display towards this. A further seven sites also had entrances which coincided with the majority of the visible pathways, however in these cases their entrances were not elaborate.

On the other hand, within the relatively subtle landscapes of both the Gower and Warminster, the sites are separated by low-lying featureless land (Figures 69 and 74). The 215

A GIS-based Analysis of Hillfort Location and Morphology sites were constructed from a preconceived plan. The multiple instances of oblong hillforts across Scotland, two of which are within the Aberdeenshire test area, indicates that there is a clear exchange of ideas across this region. Constructing sites with similar morphologies and topographical relationships, that are also visible from afar, implies that they may have reinforced a sense of identity within the disparate communities across the region.

sites themselves are on hills, which lack visual and physical prominence and their morphology, particularly the Warminster group, strongly indicates that ostentatious forms of enclosure were used to enhance the visual and physical prominence of locations. At The Bulwark, the extensive southern enclosing works were placed in its most visible area from the sites on Hardings Down (Figures 106-108). Hardings Down West enclosure also displayed its most extensive enclosing works towards neighbouring sites (Figures 160-161). Similarly, the southern enclosing works at Battlesbury were placed in its most visible area from both the wider surrounding landscape and nearby Scratchbury (Figures 101 and 103).

Whilst the oblong inner forts of Tap o’Noth and Dunnideer imply that topography was used to enhance their visual prominence, their outer forts and several other sites within this study formed subtle enclosures that strongly adhered to the topography. Due to their topographical adherence, they did not form imposing structures and these examples are found across several test areas. For example, Tap o’Noth and Dunnideer are sited in an upland and deeply incised topographical region, the outer forts at both of these sites encircle the lower slopes of the hill on which their inner enclosures are situated. In doing so they closely follow the shape of the hillside. Field photography illustrated that these outer enclosures were not visually prominent (Figures 140 and 135). Viewshed analysis also found that they were not notably more visible than any other morphological component on the sites (Figure 139 and Figure 134).

The variable nature and motivations behind the disproportionate allocation of resources within the construction of hillforts further demonstrates the variability in this class of monument. Variability in the morphological form and location of hillforts on a chronological basis was identified by Hamilton and Manley in south-east England (2001). Even though the hillforts within this book do not possess dating evidence that would enable the clarification of whether the differences in their placement, morphology and visual characteristics were as a result of chronology, variability in the directionality of the hillforts has nevertheless been identified that is mirrored in studies elsewhere.

The enclosing banks of Prestonbury Castle also follow the gentle slope of Prestonbury Common (Figure 63) which is sharply cut by the gorge of the River Teign. In the west and south, where the slopes of the gorge are at their steepest, the enclosing works are minimal. The remainder of the enclosing works face where the site is not naturally defined or defended by great slopes, however today these enclosing works are not ostentatious. They physically define an area, but they do not restrict visibility and they fail to impose upon anyone who is on the outside of them. They form both a physical demarcation of space and a symbolic disconnection between the site, and the remaining hilltop.

Topographic and morphological prominence Whilst the construction of disproportionate vallation sequences was a conscious act that was aimed at influencing the perception of those who approached, many other sites were located in such a way that their relationship with the topography made them visually prominent. For example, Tap o’Noth’ s inner fort sits slightly offset from the hilltop. However, the fieldwork photography demonstrated that from the majority of the landscape the site looks like it is highly prominent and occupies the whole summit, but from the south-east its true offset and less dominant position is visible (Figure 140). Although the inner bank is more visually prominent than the outer, viewshed analysis demonstrates that the inner fort is only very slightly more visible than the outer (Figure 139). The inner fort’s lack of adherence to the topography, in comparison to the outer, did not increase its visibility, but it did increase its visual prominence.

Whilst the enclosing works of these sites were not physically or visually prominent, the topographic locations on which they were situated were. In these cases, it was difficult to effectively distinguish between the visual prominence of the hillfort from the visual prominence of the topography. Using the natural prominence of the topography decreases the investment required to make a visually prominent site. This may have been a conscious attempt to save resources or it may simply have been that there were not sufficient resources available to construct anything else. However, it is likely that such prominent locations were highly sought after by those who were constructing these monuments, therefore competition for them is highly probable. If power was needed to secure these locations and that was reflected in access to resources, such as man power, then it is unlikely that those who secured these locations were lacking resources, even if they failed to ostentatiously portray such wealth through extravagant enclosing works.

Positioning enclosing works on the summit of a topographically prominent landform created a visually prominent enclosure. However, the intimate use of the topography meant that there was not as great a requirement for multiple, large-scale enclosing works on Tap o’Noth’s inner fort. Hillforts whose morphologies factor in the natural properties of the landscape were constructed by people who had a strong awareness of the topographical qualities of the landscape. The conscious act of positioning the inner fort slightly offset from the summit at Tap o’Noth and the creation of oblong forts at both Tap o’Noth and Dunnideer, implies that these 216

Conclusion Visual prominence and the definition of hillforts

This book has also laid question to the notion of ‘local dominance’ in terms of site images and visibility. A hillfort cannot be defined as locally dominant or locally visually strong because there is variability within that locality. The fact that the Atlas project adopts a multiple point system reduces the scope for critique based upon definition and widens the span of the project. The variability in the local dominance of hillforts and their morphological components may have been a conscious attempt to influence people’s perception of the sites through the manipulation of the site’s image to those outside of it. The manipulation of site images through disproportionate vallation sequences is explored further below.

Above it was highlighted that the visual prominence of the hillforts within this study was variable, in some cases it was caused by the actual prominence of the hillfort enclosure whilst in others it came from the topographical prominence of the landform on which they are situated. Visual prominence is considered by many to be a key defining factor of hillforts. For example, Driver argued that their visually prominent components acted as monumental and symbolic indicators of importance (2007). Speaking very broadly about the situation in Wessex, Sharples argued that visual prominence set a hillfort aside from an enclosure (2010). However, just as the means by which a location is visually prominent is variable, the nature of visual prominence is also highly variable and the fragmentary nature of this was highlighted by the results of the GIS-based analysis.

The falsification of images through disproportionate vallation sequences As it was highlighted above, several sites had their most elaborately enclosed aspects within their most accessible or visible area and their relationship with the topography, in some cases, distorted their image. However, several other examples were found where their elaborate architectural components coincided with the blind pathways. This meant that by definition, the façades of Castell Tregaron (Figure 55), Trecoll (Figure 56), Pen y Bannau (Figure 54), Cairnmore (Figure 191) and Hardings Down West Enclosure (Figure 190) were placed where there was limited visibility of the sites.

Visual prominence and overall visibility can be fragmentary, for example when additional banks were constructed within sectors that they were not defensively needed, such as at The Bulwark and Battlesbury. In some cases, hillforts that were very close together had limited intervisibility, for example the Hardings Down group, Dartmoor hillforts, and Scratchbury and Battlesbury. Other sites were highly visible from the surrounding landscape; however, the degree to which they were visible varied with distance and directionality.

Unlike the disproportionate vallation sequences at Prestonbury, The Bulwark and Battlesbury, the visual impact of the façades, which coincided with the majority of the blind pathways, was limited to those that were within the immediate vicinity of these sites, and more than likely approaching them. O’Driscoll argued that the visibility of the enclosing works from a distance may have reaffirmed the community relations of those who constructed them (2017). However, in the case of the sites with blind façades, this was not possible. Here, the focus seems to have been on influencing the first impressions of the site for those who were approaching. These façades manipulated initial impressions, they informed people’s experiences, and were part of a performance.

The highly variable visual prominence of hillforts raises the question as to whether this characteristic can be used to define a hillfort, as Sharples suggested (2010). At many sites within this study, what was visually prominent from one viewing point was not from another. Consequently, just as the term hillfort covers a wide variety of sites, the means by which a site can be defined as visually prominent is also highly variable. Visual prominence is also affected by distance, direction and vegetation growth. The variability of visual prominence further questions how one can define a site as a ‘hillfort’ compared to defining it as an ‘enclosure’. Whilst this study was in its initial stages, so too was the Atlas of Hillforts in Britain and Ireland project. During their initial discussions, the difficulty of what to include or exclude within the study was highlighted. It was noted that due to the variability of sites it would be impossible to have a single definition of what is a hillfort. Consequently, the project adopted a point system whereby for a site to be included within the database it had to satisfy two of the three criteria, these were:

Although they could not reaffirm the identity of the disparate groups within the wider landscape, and they influenced first impressions of the site, they would not always have been seen for the first time. These structures were potentially constructed by a large group of people that could have come from both within the physical community of the hillfort and its wider surroundings. Consequently, these façades were not necessarily a ‘surprise’ to everyone who approached.

1. Landscape setting and some notion of local ‘prominence’ . . . 2. Scale of enclosing works, includes entrances, something bigger than you would expect for a farmstead 3. Enclosed area, notionally 0.2ha is a boundary.

The coincidence of hillfort façades with blind pathways does not support the study’s initial hypothesis that image would be portrayed through a hillfort’s most physically or visually accessible area. In these instances, whilst efforts had been made to influence the perception of these sites through morphology, and there was a selective allocation

(Atlas of Hillforts in Britain and Ireland Project 2015) 217

A GIS-based Analysis of Hillfort Location and Morphology defensive perspective. Even if the interior was visible from a distance, visual clarity is likely to have been poor, therefore there would be limited access to information.

of resources, the selectivity of this investment and the investment in itself made a minimal impact on the wider landscape. The subtlety of these relationships and the direct nature of the disproportionate vallation sequences demonstrates that they were constructed by people who had a strong understanding of the characteristics of the underlying topography and how people moved through the landscape. Whilst this morphological investment was not as readily visible as in other sites, it still conveyed the same principle of forming part of a performance to display the prowess of these enclosing works, the resources consumed and the people that constructed them.

Although activities within the sites are not likely to have been clearly distinguishable, the lack of an impenetrable visual barrier, such as substantial enclosing works, meant that they were visually open and, to a certain degree, public monuments. Even though their enclosure physically and symbolically separated them from the surrounding landscape, the high visibility of their interior visually integrated them into it. In these cases, the enclosing works were used to monumentalise topographical features. Hamilton and Manley identified the use of enclosing works to monumentalise topography and spatially divide sites in their study of hillforts in south-east England (2001). The creation of visually open sites monumentalised a topographic location, it did not form a site that was impenetrable to those outside of it. This indicates that whilst physical access may have been restricted into these hillforts by their physical enclosure, they were visually open to varying degrees which brought them into being and made them accessible to those within the landscape.

The variable visual qualities of a hillfort’s architectural components also has an impact upon their total visual accessibility. This book has identified that the completeness of the image portrayed by hillforts, i.e. the amount of it that was visible, was highly variable. Less than 25% of the sites within this study portrayed a complete image, the remainder had split images. This demonstrates that just as interpretations of visual prominence can vary with distance and direction, the visibility of a site’s architectural components also varies. This variability demonstrates that access to knowledge is also highly variable, consequently in many respects those that are seen as ‘public monuments’ cannot necessarily be interpreted as ‘completely’ public. The variability in the degree to which hillforts formed public monuments is explored further below with a discussion surrounding the evidence for visually open and closed hillforts.

The test areas also comprised visually closed sites, however the degree to which they were visually closed is highly variable. For example, in the Dartmoor test area, there is poor intervisibility between Cranbrook Castle and Prestonbury Castle which are situated on opposite sides of the Teign River Valley. There is also poor intervisibility between Battlesbury and Scratchbury in the Warminster test area. Both of these sets of sites have been argued to constitute a pair (Griffith and Wilkes 2011, Corney and Payne 2006). In both these cases, only their outer perimeters were intervisible. In the Dartmoor example, the sectors of the sites which were visible from each other are minimally enclosed which is the basis to Griffith and Wilkes’ argument that they constitute a pair (2011). In Warminster the evidence to support the idea that Battlesbury and Scratchbury formed a pair of hillforts is not as strong as although the side of Battlesbury that is visible from Scratchbury is not the site’s most enclosed aspect, it is still highly defined by enclosing works.

Visually open and closed hillforts Whilst the enclosing works of sites were of variable visibility, the visibility of their interiors was also highly variable. A dominant idea surrounding hillforts is that they were defensive structures (Simpson 1964; Sutton 1966; Dyer 1992; Hogg 1979; Bowden 2011), consequently, it was argued that they could not function as such if their interiors were visible. This study’s GIS-based analysis found that regardless of the height of the enclosing banks at some of these sites, several were visually open. This was demonstrated at hillslope enclosures whereby their interiors were highly visible, but it was unsurprisingly highly directional. For example, on the Gower, the West and North enclosures on Hardings Down were highly visible from the lowlands to the north and west due to the slopes of the hill (Figures 160 and 148). This is mirrored in the Dartmoor test area at Prestonbury Castle which is highly visible from Broadmoor Common (Figure 86). The slope of the hill meant that the height of the enclosing banks would not have affected the visibility of the site’s interior.

Based upon the observations made within the Dartmoor and Warminster test areas, evidence for morphological directionality at the hillforts on the Gower indicates that there may have been a similar situation. Whilst the West and North enclosures on Hardings Down were visually open to the surrounding landscape, these sites alongside the East enclosure were visually closed off to each other as there is poor intervisibility (Figures 149-150, 161-163). Their entrances were also oriented away from each other which supports the observation that they were visually and physically oriented away from one another.

Even though from the modern perspective these sites were not defensible, Finney argues that from a later Prehistoric perspective this was not the case (2005). He highlights that the maximum distance that a projectile can reach is 100m, which indicates that the high visibility of a hillfort’s interior from long distances is not significant from a

The potential pairing and grouping of hillforts has been postulated elsewhere, for example the Baltinglass hillforts. Here there have been two interpretations, one saw the sites to have been occupied by competing groups, the 218

Conclusion other argued that they represent the sequential activity of one group (Waddell 1998). In the case of this study, it is unlikely that Prestonbury and Cranbrook were occupied by competing groups as the sides that face one and other are of limited enclosure. This meant that, in this case, there was no physical evidence of competitive display. It is quite possible that Prestonbury and Cranbrook represent the successive occupation of an area, or in fact that they were the same social group who occupied both sides of the valley. Scratchbury and Battlesbury could also represent successive occupations of the same topographic region. Similarly, at Hardings Down the sites occupy different sides of the hilltop which meant that, as a group, they had a good degree of visibility to the wider landscape. It is therefore possible that they were constructed sequentially to ensure maximum access and exposure to this landscape. However, whilst there is currently no absolute dating evidence for these sites, the capacity of this landscape to resource the construction of these sites also needs to be considered. Afterall, a considerable amount of resources both in man power and timber would have been required to construct them. This raises the point that it may have been difficult for these sites to exist as contemporary pairs due to the requirement for such a volume of resources, conversely their existence as a pair or group may have been an extravagant display of power towards the surrounding landscape and those that were moving through it. The following section explores further how people’s visual and physical experience of the morphological components of these hillforts was affected through movement. The concept of correct pathways is also explored.

outer fort (Halliday 1999). Whilst there is no definitive proof that these breaks were actual entrances, the break in the north-western area coincided with several GISbased pathways. Although this was one of the site’s most accessible points; it is also one of its least occupied areas. The lack of structures within this area may have been an intention of those who constructed the site as it meant that the entry point was not overcrowded, and movement would have been freer.

Correct pathways

Even though there is a strong desire to perform through architecture, there are still instances whereby their morphology was oriented to satisfy a practical and functional need. For example, in half of the test hillforts at least one of their entrances coincided with the slope-based least cost pathways. These entrances enabled movement both in their physical form, but also in their location which ensured that accessibility was easiest. Positioning entrances in locations where entry required the least amount of effort implies that access was promoted to a certain degree.

Thirteen of the test hillforts have entrances that were enhanced beyond the necessary form of a simple gap to mark an entry point. For example, at The Bulwark, the eastern entrance has outworks that enlarge the site’s footprint without expanding its occupational area (Figure 70). These earthworks did not control movement into the site, they enhanced its image to those who were approaching and entering it from the edge of the topographically prominent Llanmadoc Hill. On the other hand, at Wooston Castle the site’s entrance was approached by a hollow way that formalised movement into the site (Figure 62). This hollow way also strongly coincided with slope-based pathways (Figure 172), which demonstrates that those who constructed it enhanced the topographical qualities of the land. This enhancement satisfied the hillfort builder’s desire to accentuate moving from ‘outside’ to ‘inside’. The experiential and morphological enhancement of entry and exit into these sites further supports this study’s findings that although hillforts are superficially highly variable, there is an underlying and consistent desire to create a performance.

Another aspect of Driver’s research drew out the concept of correct paths of movement (2005a and 2013). He noted that at Pen Dinas Elerch and Castell Grogwynion direct access to the sites was obstructed by their morphologies (2013). This meant that people were routed into them through designed routeways. This GIS-based analysis investigated whether other forms of correct paths of movement existed, for example ones that were informed by visibility and ease of movement. By definition, the GIS slope-based pathways followed the least effort route which does not factor in deflection from such a route. Driver’s morphological correct pathways were constructed, actual pathways that also imply regularity in movement but in this case it was enforced regularity.

However, whilst there was evidence for influencing both the perception of the sites and promoting access, cost surface analysis indicated that at Battlesbury the north-western entrance was placed with a total disregard for enabling easy access. This was positioned offset from the plateau, which is the most accessible point of the hillfort, and this plateau also housed the Battlesbury Bowl settlement (Ellis et al. 2008). The placement of the entrance slightly offset from this plateau increases the difficulty in entering the site. This implies that accessibility was not the foremost influential factor within the placement of the entrance; it may have been selected for more symbolic reasons. The author believes that this symbolic reasoning may have been based on the need for some fort constructions to disassociate themselves from the past; after all it is believed that the Bowl settlement was the predecessor to

The degree to which this study’s sites exhibited evidence for correct pathways was variable. On a morphological basis, the inner fort at Tap o’Noth does not exhibit any evidence for an entrance. However, there is a high correlation of slope-based least cost pathways with its highly eroded south-eastern corner (Figure 174), which could imply that it was a natural entry point at some stage. There are also several breaks in the enclosure circuit of the outer fort, some of which are believed to be potentially ‘ancient’ (Halliday 1999; Halliday 2008). These are believed to have been access points to the huts within the 219

A GIS-based Analysis of Hillfort Location and Morphology It raises the question as to whether all locations which had good accessibility, access to resources and the past could be considered ‘prime’ (Driver 2005a; Driver 2013), or whether one drew people more than the other. Good accessibility encourages movement between hillforts and their surrounding landscape; it allows resources such as food and material into and out of the site to maintain a community. Associations with the past can imbue a location with importance, the continued investment at these locations with new additions potentially inherit this importance. For example, The Bulwark was situated on the eastern edge of Llanmadoc Hill, which is densely occupied with Bronze Age barrows. This site’s morphology implies that although those who chose to construct The Bulwark wished to be associated with this significant place, its association was limited as the most prominently enclosed aspects of the site face this area. The eastern entrance also faces away from it. This meant that physical access to the remainder of Llanmadoc Hill was made more difficult than it could have been. Viewshed and fieldwork observations also found that there was limited intervisibility between the remainder of Llanmadoc Hill and The Bulwark. A similar situation also occurred on Hardings Down. Here, the sites’ entrances are all situated away from one another. Cost surface analysis demonstrated that their entrances were not placed where the sites were most accessible from each other (Figures 177-178). This increased the complexity of travelling between them.

the hillfort. The degree to which sites and features from the past influenced the morphology of the test hillforts and movement within the landscape is explored below. Beyond movement, visibility and topography Physical associations with the past Whilst constructing hillforts embodied ideas that were present at the time that they were constructed, it was also affected by associations with the past (O’Connor 2009, 11). Driver brought forward the concept of prime and non-prime locations which were primarily defined on the basis of their accessibility, but were also related to the past (2005a). He argued that when sites were physically associated with evidence from the past it indicated that the location had been important for a long time (Ibid.). Gosden and Lock investigated the idea of ‘Prehistoric Histories’ by exploring the concepts of history and myth (1998). They argued that ‘history and myth are not polar opposites, but are mutually linked’ (Ibid., 11). People were aware of the past, however how that related to, and how it was used within the present, potentially differed across time (Ibid. 1998) and space. This GIS-based analysis identified several instances where hillforts were constructed in established places of importance, however how they related to this differed greatly. In the majority of cases, where sites were physically associated with evidence of previous use, it often physically hindered site construction as material had to be derived from further afield than from the immediate hilltop. For example, in the case of Battlesbury the preservation of the barrow in the site’s southern corner meant that it was incorporated within the defences (Figure 76). This increased the difficulty in constructing the southern bank. This barrow is also located within one of the site’s most visible areas; consequently, the contrasting form of this barrow with the relatively flat interior meant that it potentially detracted attention from the hillfort itself.

The variable nature of relationships between the test hillforts and other elements of prehistoric activity in these landscapes further demonstrates the lack of consistency in their placement and morphology. There was not one dominant morphological response to neighbouring hillforts and non-hillfort sites. However, this as with other observations, needs to be set within a wider chronological context, where possible. Wider chronological picture The construction of hillforts came at a time when there was an increased emphasis on the definition of space (Hingley 1990; Lock 2011), which had its origins in the Neolithic (Hill 1996). This study identified sites whose morphology and location implied that they served to define, secure and dominate space. However, just as the blanket term ‘hillfort’ covers a wide variety of sites, the means by which these sites defined, secured and dominated spaces also varied. Such variability may have been enhanced by topography, which was raised earlier within this chapter; however, it may also have been chronological.

The position of barrows within hillforts also hindered intra-visibility at them. This was demonstrated at Cley Hill, which has two very visually prominent barrows in its centre (Figures 75). Just like at Battlesbury these barrows detract attention from the hillfort due to their visual prominence. In these cases, the landscape features from the past, by being preserved within the hillforts, were allowed to visually and physically dominate these more recent landscape investments. This demonstrates the importance that the past was given at these sites, an importance that was acknowledged and incorporated within the present. Associations with the past may have enhanced the power of the communities which associated themselves with these locations.

There is little definitive chronological understanding of the development of hillforts, particularly their morphologies. It is broadly believed that within the Middle Iron Age (400-300 BC) many early hillforts were abandoned in central southern Britain and the remainder were refortified (Cunliffe 2006). These middle Iron Age hillforts developed from sites that were enclosed with a singular bank and ditch (Ibid.) to more morphologically complex structures.

There are also instances where sites completely disassociated themselves from the past, and in some cases their present. This was particularly seen on The Gower. 220

Conclusion Cranbrook Castle and Castell Tregaron occupy rounded hilltops, as hillforts often do; however, the relationship between their morphologies and the topography indicates that the sites were superimposed upon the topography. Their morphology was applied to, and essentially cut into, the topography in a very abrupt form, whereas Wheedlemont and Cairnmore did not. Even though the morphology of Cairnmore and Wheedlemont was not elaborate, their enclosure was relatively extensive, compared to their small size (Gondek and Noble 2011). This led to speculation, prior to the dating of Cairnmore, that they could have been occupied within the early Medieval period (Ibid.). These early Medieval sites largely followed the topographical form of the hilltop, and whilst they responded to the susceptibility of the site it was not ostentatiously done, which mirrors what occurred at a number of the presumably prehistoric sites. Consequently, morphology and location are not good dating guides.

To some extent, this chronological pattern could be seen within the small sample of Britain’s hillforts, that were investigated within this study. For example, at Warminster it is believed that because Scratchbury was morphologically very simple and poorly defended, in comparison to Battlesbury, it was earlier (McOmish, Field et al. 2002; Ellis, Powell et al. 2008). Wooston Castle is also believed to have increased in morphological complexity over time, however Historic England argue that this site was not established until the 1st and 2nd centuries BC (2015b), which does not support Cunliffe’s model (2006). The morphologies of Dunnideer and Tap o’Noth did not develop in the same way as those in the Warminster test area. Instead, it is believed that the oblong forts were of a later phase than the outer forts (Cook 2013a; Feachem 1966). These oblong forts were visually and physically impenetrable however their enclosing walls particularly at Tap o’Noth were visually and physically imposing. The addition of these oblong structures to the outer fort implies changing values within society. These prominent inner forts formed the public face to the sites which had earlier been relatively subtle constructions on prominent topographical features. As Lock, Gosden and Daly argued, constructing hillforts drew upon the past whilst also reassembled it within ‘new relationships of the present’ (2005, 134). Constructing hillforts was a ‘community project. . .that. . . helps bring a community into being’ (Ibid.). Lock and colleagues argued that the construction of a new site such as Segsbury created a ‘novel community’ (2005, 134). The construction of the inner forts at Dunnideer and Tap o’Noth which were such a contrast to their predecessors implies that this was also a ‘novel community’.

Concluding remarks Even though this study identified a great deal of variability behind the location and morphology of hillforts, there is an overarching similarity. There is an overarching drive in hillfort construction to employ directionality in their morphologies. This directionality focuses upon influencing people’s experiences of them and it went beyond pure practicality. As with Mitcham’s study of Hampshire hillforts (2002), this book did not identify sites that were overtly defensive in form and siting, their primary focus was that of display. Whilst at some point some hillforts may have been constructed with defence in mind, the visual and physical properties of these structures appear to be highly social.

Cook postulated that Tap o’Noth could have been reused within the Early Medieval period (2011). This site towers over Rhynie, which has a dense concentration of activity associated with the Early Medieval period. However, as Ralston highlighted, it is unlikely that the inner fort at Tap o’Noth had any substantial occupation within this period as there are still upstanding traces of the enclosure which predates the inner fort itself (2004).

Whilst this book focused purely upon the analysis of the visual and physical properties of sites, there is the opportunity for future studies to further explore the visual and physical qualities of hillforts compared to random locations. This could further our understanding of the reasoning behind hillfort placement and develops upon the work of Marsh and Schreiber (2015). Similarly, visual relationships between hillforts and the surrounding landscape should also be explored, for example to and from other sites.

The early Medieval forts of Wheedlemont and Cairnmore had minimal visual and physical prominence within the surrounding landscape (Figures 128, 131). The visual and physical prominence of these sites has inevitably been affected by the degree to which they survive today. Even though these sites are likely to be heavily eroded, the degree of this erosion would never have meant that they were as visually prominent as similar sized hillforts within similar topographical points such as Cranbrook Castle and Castell Tregaron. The date of these sites is not known; however, it is assumed that they derive from the later prehistoric period.

However, without a true chronological understanding of the development of hillforts it will always be difficult to ascertain the depth of the visual and physical relationships within these landscapes. It is imperative that we prioritise investigating the chronology of hillforts on a national scale. Establishing a chronological understanding of hillforts would greatly enhance exploratory analyses like this. The following chapter explores the effectivity of this study’s methodology and highlights how it can be developed upon for future research.

221

6 GIS Approaches and British Hillforts – Lessons Learnt and Future Prospects however found to be minimal. Although the effect of differing DTMs was minimal, if the study was to be done again, with more time, and if the LiDAR datasets were complete, then the methodology should be considered to be applied to a LiDAR based DTM. However, as it was highlighted in Chapter 2 the use of LiDAR needs to be taken with caution as high resolution processed LiDAR DTMs can have inherent artefacts. Although, when used with care, LiDAR based DTMs would still enable the analysis of much subtler relationships on a landscape scale particularly with regard to least cost pathways. LiDAR will also provide a greater scope to assess the individual morphological components of sites. It also enables experimentation with the manipulation of site morphologies to factor in buildings and other architectural features that could have affected visibility.

This book set out to develop a GIS-based methodology to investigate the degree to which the construction and morphology of a selection of hillforts across Britain enabled or disabled an individual’s ability to gain visual and physical knowledge of them whilst being within the landscape. The focus of this work was based upon, and developed from the work of Driver who advocates that hillforts need to be examined as ‘three-dimensional ‘artefacts’’ by assessing them in relation to human experience and the approaches to them (2007, 86). The innovative approaches applied in this GIS-based methodology, with the integration of movement and visibility alongside the utilisation of radii in both Cost Surface and Viewshed Analysis quantifies and enhances one’s understanding of the potential to experience hillforts. The addition of ground-truthing field visits also meant that the scope of the GIS-based results could also be tested and enhanced humanistically.

From the outset it was acknowledged that the analysis of the Aberdeenshire test area without any LiDAR data would be difficult, however the extent of that difficulty was underestimated. Without LiDAR data the analysis of the Aberdeenshire test area was reliant upon acquiring detailed plans of the sites. These subsequently had to be georeferenced within ArcGIS which greatly increased the processing time for the test area. The aim of creating this methodology was to not only answer analytical questions but to also harness the ability of GIS to analyse large quantities of data within a short period of time. A test area without LiDAR data severely reduces the speed and effectiveness of using GIS for analysis. This detracts from its speedy and convenient nature.

Revisiting the aims and objectives As with every piece of research, this study set out with a series of aims and objectives. It had two fundamental aims: • To develop a GIS-based methodology to investigate the morphology and location of test hillforts. Focus was put upon the assessment of movement and visibility in relation to hillfort location and morphology. • To test the results of Driver’s non-GIS-based analysis of hillfort location and morphology (2005a and 2013) with a GIS-based analysis.

Although this study was primarily GIS-based, in planning the methodology the need to gain an experiential understanding of the hillforts within their landscapes could not be ignored. The importance of experientially understanding hillforts was emphasised and reinforced through Driver’s innovative approach to hillfort location and morphology (2005a; 2007; 2013). GIS-based analysis provides a measured and objective approach to the analysis of hillfort location and morphology. However, these sites also need to be understood by seeing them within their landscape. Site and landscape visits allow one to gain an understanding of their visual prominence, which a GIS cannot effectively depict. A GIS provides a quantitative analytical environment. It is also a visualisation tool that provides what is essentially a bird’s eye view of the landscape; it cannot immerse people within a landscape as being within one can, as this gives you the threedimensional understanding of a site’s location and morphology. Visual studies based upon GIS have been criticised as they ‘[reduce] the perceptual act. . . to the delineation of a viewshed, which is in turn presented as an

A critical analysis of the methodology and data A methodology was created that enabled the author to investigate the visibility and physical accessibility of the test hillforts. Whilst the methodology was productive in allowing effective interpretations of the results, limitations with the data and methodology were highlighted. These can be improved in future applications. From the outset the writer highlighted that caution was needed when using GIS-based analysis, as it can be very deterministic. Efforts were made throughout to counteract this effect by using multiple DTMs with different resolutions and multiple weightings of the friction surfaces. Field visits were also used to ground truth the results. The primary base map DTM for the test areas was Ordnance Survey PROFILE DTM, however small sub-areas of the test landscapes were also analysed using a LiDAR based DTM. This exercise enabled a comparison of the results from two different DTMs, however it was only undertaken when possible. The differences between the results were 222

GIS Approaches and British Hillforts – Lessons Learnt and Future Prospects the banded viewsheds from the radii and hillfort grids identified which aspects of the site were most visible and from where. For example, the southern enclosing works of The Bulwark are highly visible from Hardings Down as opposed to the wider landscape (Figures 106–111).

aseptic zone on a flat projected mapsheet’ (Wheatley and Gillings 2000, 13). Field visits add a qualitative aspect to an otherwise quantitative investigation of movement and visibility. Does the application of GIS-based analysis enhance Driver’s non-GIS-based analysis?

However, the degree to which the results of the GIS-based investigation of image compared to the non-GIS-based analyses of it were variable. The GIS-based methodology enabled the identification of how areas of extensive enclosure related to the accessible and visible nature of the area in which they sit, which was not easily done through site visits. This is likely the cause as to why there is a degree of disagreement between Driver’s interpretation of a Ceredigion site’s image compared to the quantified results of the GIS-based investigation into the same site.

GIS-based analysis was applied to a subset of Driver’s test hillforts, and to the investigation into three of Driver’s key observations of: • The physical relationship between hillfort morphology and topography- this focused upon analysing whether topography was used to aid the construction of the test hillforts, or whether it was disregarded in the construction process and a ‘design’ was essentially placed upon the landscape. • Image – Image was defined by the disproportionate allocation of resources within the construction of hillforts in places which were exposed to large numbers of people (for example an area that was highly visible or where it was close to a routeway and consequently visually or physically accessed by a large number of people) • Correct pathways – in terms of Driver’s non-GISbased analysis, these were defined either in terms of topography i.e. a topographically defined routeway or on a morphological basis through a complex entrance arrangement. These morphological correct pathways could have been designed to reinforce or manipulate the image of a site. The current GIS-based analysis investigated how the GIS-based correct pathways as defined by least cost pathways (slope, visible and blind) corresponded with Driver’s concept of a correct pathway.

Visual prominence is not as measurable with a GIS, but it is interpretable in the field. However, this interpretation is highly subjective and affected by the experiences of the observer. In the case of Driver’s observations of his test hillforts and the author’s own observations at hillforts within her test areas, the judgement of visual prominence was inevitably influenced by subconscious comparisons between the sites under question and those sites that the author had previously visited. This subjectivity inevitably influenced the overall conclusions of studies, particularly those like Driver’s who are reliant upon non GIS-based analysis (2005a and 2013). However, the combination of GIS-based analysis with fieldwork observations meant that the possibility of such subjectivity placing a bias upon the study’s conclusions was reduced within this book. The data from the quantitative GIS-basis of this study allows one to have a measurable understanding of the visual and physical qualities of sites. This also provides the study with precision and comparability, which is otherwise not readily possible within non-quantitative analysis.

The physical relationship between hillfort morphology and topography

Future applications of this methodology to the investigation of image will be enhanced through further GIS-based analysis. This could include the application of visual prominence calculations such as those adopted by Llobera (2003). Llobera investigated both topographical and visual prominence within different neighbourhood scales (Ibid.). He defined visually prominent locations as those which are most visible compared to the remainder of the neighbourhood. Whilst topographical prominence was based upon the morphological character of locations within a neighbourhood. This approach is highly applicable to an investigation of hillforts as it will provide an insight into how the visual and topographical prominence of hillfort locations compared to their surrounding landscape and the impact of such prominence.

LiDAR enabled the analysis of hillfort location and morphology in relation to the wider topography. This identified how topography was incorporated in the morphology of test hillforts. The results of this analysis supported those of Driver’s in Ceredigion (2005a; 2013; Chapter 3); it also strengthened and reinforced the discussions in other test areas. However, due to the lack of available LiDAR data for the Aberdeenshire test area this analysis proved to be very difficult here. It was still possible to assess the physical relationship between site morphology and topography; however, it was not possible to clarify the finer details of the relationships as it was in the test areas with LiDAR coverage.

Further means by which this methodology can be enhanced include applying a quantitative analysis of how distance and visual acuity affects visual clarity within a landscape, similar to Ogburn’s approach (2006). Whilst this book examined changes in site images with visibility studies from the buffer zones and the neighbouring hillforts, the

Image The visibility of the morphological components of sites such as the most extensive enclosing works, or the entrances, was investigated through viewshed analysis. As the earlier chapters within this book demonstrated, 223

A GIS-based Analysis of Hillfort Location and Morphology indicated that at least one of the entrances to each of these sites was placed in the site’s most accessible area (Figures 172, 173, 176). However, it failed to support Driver’s morphologically defined correct pathway at Castell Grogwynion (Figure 38).

methodology could be further enhanced by increasing the extent of viewshed analysis. This approach would see the computation of viewsheds along known and GIS generated pathways. The computation of viewsheds along routeways has already been applied by Chapman, whose results demonstrate the merit in such a study (2000 and 2003). Viewsheds could also be generated at potential waypoints/obstacles within the landscape such as barrows and settlements.

There was very little agreement between Driver’s correct pathways for Ceredigion and those that were generated within the GIS. Driver’s correct pathways were topographically and morphologically defined, they were both natural and constructed pathways. GIS generated least cost pathways, by definition, follow the least effort routes through the landscape, which were not necessarily the case with Driver’s pathways. Least cost pathways also have a set origin and destination, they imply regularity in movement which was not necessarily the case. This is a potential explanation as to why there is so little correlation between Driver’s correct pathways and those generated within the GIS. The value in the GIS generated least cost pathways lies in their potential to identify why routeways/entrances were placed in such a way by comparing routes of least cost pathways to known routes such as hollow ways, braided trackways and activity areas. This approach follows on from studies such as Symonds and Ling (2002), White and Barber (2012), and Güimil-Fariña and Parcero-Oubiña (2015).

Correct pathways The primary methodology for investigating the concept of correct pathways was through GIS-based cost surface analysis. The results of this analysis were subsequently related to both topographical features and evidence for human activity. This enabled an investigation as to whether the least cost pathways travelled through significant places within the landscape. Areas became significant through people’s successive encounters with them, which transformed them into places (Giles 2007, 110). Investment was subsequently undertaken within these areas over successive periods. Both the results of the investment and the topographical areas in themselves, persistently drew people to these areas consequently they are likely to be an indicator of routeways (Driver 2005a).

The innovative approach of integrating viewshed results with Cost Surface Analysis allowed the investigation of how movement affected the visibility of sites. For example, in the case of Prestonbury the computation of least cost pathways based upon the visibility of the individual architectural components of the sites indicated that it largely portrayed a complete image (Figure 93). The integration of movement with visibility through the calculation of blind pathways indicated that the façades at Castell Tregaron, Trecoll, Pen y Bannau, Cairnmore and Hardings Down West enclosure were all placed to face an area where those approaching had limited visibility of the site (Figures 55, 56, 54, 190). Visible pathways in relation to Castell Grogwynion, Gaer Fawr, Cranbrook Castle, Wooston Castle, Prestonbury Castle, Dunnideer and Scratchbury all indicated that their entrances were placed where those within the site had maximum visibility to approaching traffic (Figures 47, 48, 181, 182, 183, 184). Observations like this would be difficult to make objectively in the field as one is influenced by their experiences within the landscape, GIS enables the objective computation of these pathways which can subsequently be ground-truthed.

The correlation of least cost pathways with products of human activity and topographical features was most visible in the Aberdeenshire test area. Here there was a significant correlation between the slope-based and visible least cost pathways with extant or contemporary areas of activity at the time that the hillforts were in existence (Figures 194-200). This indicates that there was activity within these locations, people were demonstrably aware of these places. Once people became aware of them it is likely that they became part of their consciousness and subsequently part of their world of movement. These results also imply that the GIS-based pathways may indeed relate to actual pathways. However, the overall poor coincidence of least cost pathways with known routes in the majority of test areas demonstrates that movement both now and in the past was far more complex than following the least effort route. Even though the results of these comparisons predominantly failed to confirm that the least cost pathways followed actual routes, this approach was still current as it aligned itself with current foci of GIS-based studies (GüimilFariña and Parcero-Oubiña 2015; White and Barber 2012). Its computation of slope-based least cost pathways to and from the hillfort buffers also meant that it aligned itself with studies that use the ‘from everywhere to everywhere model’ (White and Barber 2012; Nolan and Cook 2012).

Whilst this GIS-based methodology focused upon physical constraints during the computation of least cost pathways, movement in the past is also likely to have been influenced by cultural factors. Consequently, further applications of this methodology will be enhanced through an investigation of how cultural factors could also influence the routes of least cost pathways. Factors such as the avoidance of field systems, non-destination/ non-origin settlement areas and ceremonial landscapes could be factored into friction surfaces. Similarly, whilst this work took steps in aligning itself with studies that

Slope-based Cost Surface Analysis was applied to the investigation of site accessibility and how that related to morphology. This developed on Driver’s concept of a correct pathway, which he defined topographically and morphologically (2005a and 2013). For example, slopebased Cost Surface Analysis within the Dartmoor test area 224

GIS Approaches and British Hillforts – Lessons Learnt and Future Prospects Secondly, the investigative combination of GIS-based analysis alongside field visits should be applied to a singular ‘region’ of hillforts. This region ought to be defined topographically to avoid biases within the data as a result of modern cultural and arbitrary boundaries, which did not exist in the past.

use the ‘from everywhere to everywhere’ model, this methodology could be improved by undertaking least cost pathway analysis from each grid cell based upon both slope and visual friction surfaces. The step forward and future research prospects

Finally, the extent to which this methodology can progress is limited to the quality and nature of the data that is available. It will be enhanced greatly once a better understanding of hillfort chronology, phasing, and the distribution and density of interior buildings is established. This will enable researchers to relate the visual and physical relationships on a chronological basis, whilst also investigating the sites as functioning entities that have potentially influential interior components.

This study’s innovative approach to examining movement and visibility in relation to hillfort morphology made further steps in initialising the development of hillfort studies beyond discussions of defence and typologies. Further steps must be taken to continue this progression. Firstly, the GIS-based methodology must be enhanced by incorporating factors such as visual acuity and social costs through friction surfaces. The addition of these components to this methodology will greatly progress studies by equipping them with the potential to widen the scope of investigating the reality of the GIS-based least cost pathways compared to known routeways. It will also enable studies to identify the motivations behind routeway and hillfort location alongside hillfort morphology.

In conclusion, a number of areas have been identified where this methodology can be developed, however this list is by no means exhaustive. The potential for GIS-based analysis to aid the development of a better understanding of complex monuments such as hillforts is endless.

225

Bibliography Bertin, J. Graphics and Graphic Information-Processing, Berlin: Walter de Gruyter, 1981.

Aberdeenshire Council Archaeology Service. Sites and Monuments Record extract. Aberdeen: Aberdeenshire Council, 2014.

Bongers, J., Arkush, E. and Harrower, M. “Landscapes of death: GIS-based analyses of chullpas in the western Lake Titicaca basin. “Journal of Archaeological Science 39, no.6 (2012): 1687-1693.

Anderson, J. Scotland in pagan times: the iron age. Edinburgh: David Douglas, 1883. Arbour, C. Site and scene: evaluating the context of visibility in monument placement during the Neolithic and Bronze Age of West Penwith, Cornwall, England Master of Arts, Trent University, 2011.

Bowden, M. Hillforts, English Heritage, 2011. Bowden, M. and McOmish, D. “The required barrier.” Scottish Archaeological Review 4 (1987): 76-84.

Armit, I. Celtic Scotland: Iron Age Scotland and its European context. London: B.T. Batsford, 2005.

Bradley, R. Ritual and Domestic life in Prehistoric Europe. Abingdon: Routledge, 2005.

Atlas of Hillforts in Britain and Ireland Project “An Atlas of Hillforts in Britain and Ireland” Atlas of Hillforts in Britain and Ireland, accessed 20 Dec, 2015. http:// www.arch.ox.ac.uk/hillforts-atlas.html

Bradley, R. The Prehistory of Britain and Ireland. Cambridge: Cambridge University Press, 2007. Bray, W. and Tramp, D. The penguin dictionary of archaeology. Middlesex, Penguin Books, 1970.

Atlas of Hillforts in Britain and Ireland Project. Point data for Hillforts. University of Edinburgh and University of Oxford, 2018.

Brughmans, T., Keay, S. and Earl, G. “Understanding Intersettlement Visibility in Iron Age and Roman Southern Spain with Exponential Random Graph Models for Visibility Networks.” Journal of Archaeological Method and Theory 22, no.1 (2015): 58-143.

Avery, M. “Hillforts of the British Isles: A student’s introduction.” In Hillforts later prehistoric earthworks in Britain and Ireland, edited by Dennis Harding, 1-58. London: Academic Press, 1976.

Champion, T. “Socio-economic development in Eastern England in the first millennium BC.” In Europe in the first millennium BC, edited by K. Kristiansen and J. Jensen, 125-144. Sheffield Archaeological Monograph 6, Sheffield: J.R. Collis Publications. 1994.

Avery, M. “Cashel-na-Vean, Rosguill, Co. Donegal.” Ulster Journal of Archaeology 54/55 (1991/1992): 113119. Barclay, A., Cromarty, A. M.., Gosden, C., Miles, D., Palmer, S. and Robinson, M. “The White Horse and its landscape.” In Uffington White Horse and its landscape. Investigations at White Horse Hill, Uffington, 1989-95 and Tower Hill, Ashbury, 1993-4, edited by D. Miles, S. Palmer, G. Lock, C. Gosden & A. M. Cromarty Oxford Archaeology, Thames Valley Landscapes Monograph No.18, 243-268. Oxford: Oxford University School of Archaeology, 2003.

Champion, T. “The later Bronze Age.” In The archaeology of Britain. An introduction from the Upper Palaeolithic to the Industrial Revolution, edited by J. Hunter and I. Ralston, 95-112. Oxon: Routledge, 1999. Chapman, H. “Understanding wetland archaeological landscapes: GIS, environmental analysis and landscape reconstruction; pathways and narratives.” In Beyond the map: Archaeology and spatial technologies, edited by Gary R. Lock, 49-59, Netherlands: IOS Press, 2000.

Barrett, J. C. “Chronologies of landscape.” In The archaeology and anthropology of landscape. Shaping your landscape edited by P. J. Ucko and R. Layton, 2130, London: Routledge, 1999.

Chapman, H. P. “Rudston ‘Cursus A’- Engaging with a Neolithic Monument in its Landscape Setting Using GIS.” Oxford Journal of Archaeology 22, no.4 (2003): 345-356.

Bedwin, O. “Aspects of Iron Age settlement in Sussex.” In Aspects of the Iron Age in central Southern Britain, edited by B. Cunliffe and D. Miles, 46-51, Oxford: Oxford University Committee for Archaeology, 1984.

Chapman, H. Landscape Archaeology and GIS. Gloucestershire: Tempus Publishing Limited, 2006. Chitty, L. F. “How did the hill-fort builders reach the Breiddin?” Archaeologia Cambrensis 92 (1937):129-150.

Bell, T. and Lock, G. “Topographic and cultural influences on walking the Ridgeway in later prehistoric times.” In Beyond the Map: Archaeology and Spatial technologies, edited by G. Lock, 85-116. Netherlands: IOS Press Limited, 2000.

Chitty, L. F. “Iron Age “B”- South Western culture: showing also camps with inturned entrances.” In The Personality of Britain. 3rd ed, edited by C. Fox, Cardiff: University of Wales, 1938. 226

Bibliography Clarke, D. L. “A provisional model of an Iron Age society and its settlement system.” In Models in Archaeology, edited by D.L. Clarke, 801-869, London: Methuen & Co Ltd, 1972.

Cunliffe, B. Cradle of England. London: British Broadcasting Corporation, 1972.

Clarke, D. L. Analytical Archaeology. London: Methuen & Co Ltd, 1978.

Cunliffe, B. Iron Age communities in Britain: An account of England, Scotland and Wales from the seventh century BC until the Roman conquest. 2nd ed. London: Routledge & Kegan Paul Ltd, 1978.

Collis, J. “Cranbrook Castle, Moretonhampstead, Devon. A new survey.” Proceedings of the Devon Archaeological Society 30 (1972): 216-221.

Cunliffe, B. Danebury: an Iron Age hillfort in Hampshire. The excavations, 1969-1978: the site vol.1. London: Council for British Archaeology, 1984.

Collis, J. “Cranbrook Castle revisited.” Proceedings of the Devon Archaeological Society 37 (1979): 191-194.

Cunliffe, B. Iron Age communities in Britain: An account of England, Scotland and Wales from the seventh century BC until the Roman conquest. 3rd ed. London: Routledge, 1991.

Collis, J. “Reconstructing Iron Age society.” Europe in the first millennium BC, edited by K. Kristiansen and J. Jensen, 31-39, Sheffield Archaeological Monograph, Sheffield: J.R. Collis Publications, 1994a.

Cunliffe, B. “Introduction.” In The Danebury Environs Programme: The prehistory of a Wessex Landscape vol.1. Oxford: Institute of Archaeology, 2000.

Collis, J. “The Iron Age.” In Building on the past. Papers celebrating 150 years of the Royal Archaeological Institute, edited by B. Vyner, 123-148, London: Royal Archaeological Institute, 1994b.

Cunliffe, B. “Iron Age Societies in Western Europe and Beyond, 800-140BC.” In The Oxford Illustrated History of Prehistoric Europe, edited by B. Cunliffe, 336-372, Oxford: Oxford University Press, 2001.

Colt Hoare, R. The Ancient History of Wiltshire, FinsburySquare: Lackington, Hughes, Harding, Mavor and Lepard, 1821.

Cunliffe, B. and Poole, C. Danebury an Iron Age hillfort in Hampshire. The excavations 1979-1988: the site. London: Council for British Archaeology, 1991.

Connolly, J. and Lake, M. “Putting GIS to work in archaeology.” In Geographic information systems in archaeology, edited by J. Connolly and M. Lake, 3350, Cambridge: Cambridge University Press, 2006.

Cunliffe, B. “Understanding hillforts: have we progressed?” In The Wessex Hillforts Project: Extensive survey of hillforts in Central Southern England, edited by A. Payne, M. Corney and B. Cunliffe, 151-162, London: English Heritage, 2006.

Cook, M. “New light on oblong forts: Excavations at Dunnideer, Aberdeenshire.” Proceedings of the Society of Antiquaries of Scotland 140 (2010a): 79-91.

Dartmoor National Park Historic Environment Record. Historic Environment Record extract, Bovey Tracey: Dartmoor National Park Authority, 2013.

Cook, M. “The Hillforts of Strathdon: 2004-2010.” In ‘A Lad O’Pairts’: 62-65, Kings College Conference Centre, Aberdeen University, 2010b.

Dorling, P. and Wigley, A. Combined Herefordshire & Shropshire Report for an Assessment of the Archaeological and Conservation Status of Major Later Prehistoric Enclosures in Herefordshire and Shropshire. Herefordshire Council and Shropshire Council, 2012.

Cook, M. “Romans, Picts, and Development: Continuity and Change in Aberdeenshire’s Archaeology and Informed Planning Decisions.” Stirling International Journal of Postgraduate Research 1, no. 1 (2012): 1-13.

Doyle, J. A., Garrison, T. G. and Houston, S. D. “Watchful realms: integrating GIS analysis and political history in the southern Maya lowlands.” Antiquity 86, no. 333 (2012):792-807.

Cook, M. “Open or Enclosed: Settlement Patterns and Hillfort Construction in Strathdon, Aberdeenshire, 1800 BC–AD 1000.” Proceedings of the Prehistoric Society 79, (2013): 327-352.

Driver, T. “Tan-y-Ffordd, Hillfort”, Coflein, n.d., accessed 27 July 2015, http://www.coflein.gov.uk/en/ site/303570/details/TAN-Y-FFORDD,+HILLFORT/.

Corney, M. and Payne, A. “The regional pattern.” In The Wessex Hillforts Project: Extensive survey of hillforts in Central Southern England, edited by A. Payne, M. Corney and B. Cunliffe, London: English Heritage: 131-150, 2006.

Driver, T. “Pen Dinas Hillfort, Aberystwyth”, Coflein, 2005, accessed 27 July 2015, http://www.coflein.gov. uk/en/site/92236/details/PEN+DINAS+HILLFORT,+ ABERYSTWYTH/.

Crawford, O. G. S. and Keiller, A. Wessex from the Air. Oxford: Clarendon Press, 1928.

Driver, T. G. The hillforts of north Ceredigion: Architecture, landscape approaches and cultural contexts. Doctor of Philosophy, University of Wales: Lampeter, 2005a.

Cunliffe, B. “Some aspects of hill-forts and their cultural environments.” In The Iron Age and its Hill-Forts, edited by D. Hill and M. Jesson, 53-70, Southampton: Southampton University Archaeological Society, 1971.

Driver, T. “New surveys of two Iron Age hillforts in north Ceredigion.” Archaeology in Wales 45 (2005b): 94-98. 227

A GIS-based Analysis of Hillfort Location and Morphology Driver, T. “Hillforts and human movement: Unlocking the Iron Age landscapes of mid Wales.” In Prehistoric and Roman Landscapes, edited by A. Fleming and R. Hingley, 83-100, Macclesfield: Windgather Press, 2007.

Forde-Johnston, J. Hillforts of the Iron Age in England and Wales: A survey of the surface evidence, Liverpool: Liverpool University Press, 1976. Fox, A. “South-Western Hill-Forts.” In Problems of the Iron Age in Southern Britain, edited by Sheppard.S. Frere, 35-60, London: Institute of Archaeology, 1961.

Driver, T. “Gaer Fawr; Gaer Fawr Hillfort,” Coflein, 2012. Accessed 27 July 2015. http://www.coflein.gov.uk/en/ site/303579/details/GAER+FAWR%3BGAER+FAWR +HILLFORT/.

Fraser, I. and Halliday, S. “The Early Medieval landscape.” In the shadow of Bennachie: a field archaeology of Donside, Aberdeenshire, 115-135, RCAHMS. Edinburgh, The Society of Antiquaries of Scotland, 2008.

Driver, T. Architecture, regional identity and power in the Iron Age landscapes of Mid Wales. The Hillforts of North Ceredigion. Oxford: BAR Publishing, 2013. Driver, T. “Castell Grogwynion,” Coflein, 2014. Accessed 27 July 2015, http://www.coflein.gov.uk/en/ site/303671/details/CASTELL+GROGWYNION/.

Gentles, D. S. Archaeomagnetic directional studies of large fired structures in Britain. Doctor of Philosophy Plymouth Polytechnic, 1989.

Dunwell, A. and Strachan, R. “Tap o’Noth.” Discovery and Excavation in Scotland: 11, 1997.

Gibson, J. J. The ecological approach to visual perception Boston: Houghton Mifflin Company, 1979.

Dyer, J. Hillforts of England and Wales. Buckinghamshire: Shire Publications Ltd, 1992.

Giles, M. “Refiguring rights in the early Iron Age landscapes of east Yorkshire.” In: The earlier Iron Age in Britain and the near Continent, edited by C. Haselgrove and R. Pope, 103-118, Oxford: Oxbow Books, 2007.

Dyfed Archaeological Trust. Historic Environment Record extract. Llandeilo: Dyfed Archaeological Trust, 2014. Ellis, C., Powell, A. B. and Hawkes, J. An Iron Age settlement outside Battlesbury Hillfort, Warminster, and sites along the Southern Range Road. Salisbury: Wessex Archaeology and Defence Estates. 22, 2008.

Glamorgan-Gwent Archaeological Trust. Historic Environment Record extract. Swansea, GlamorganGwent Archaeological Trust, 2013.

Environment Agency. “LiDAR composite, [Online], 1m resolution, LiDAR Digital Terrain Model,” Geomatics, 2015. Accessed 17 Feb 2015, https://www.geomaticsgroup.co.uk/geocms/.

Gondek, M. and Noble, G. “Together as one: The landscape of the symbol stones at Rhynie.” In Pictish Progress: New studies on Northern Britain in the Early Middle Ages, edited by S. T. Driscoll, J. Geddes and M. A. Hall, 281-306, Leiden, Koninklijke Brill NV, 2011.

Esri. “Using Viewshed and Observer Points for visibility analysis,” Esri, 2012. Accessed 4 Oct 2013, http:// resources.arcgis.com/en/help/main/10.1/index. html#//00q90000008n000000 .

Gosden, C. and Lock, G. “Prehistoric histories.” World Archaeology, 30 (1998): 2-12. Griffith, F. M. and Wilkes, E. M. “In the footsteps of pioneering women: Some recent work on Devon hillforts.” In Recent archaeological work in SouthWestern Britain. Papers in honour of Henrietta Quinnel, edited by S. Pearce, 121-137, Oxford, BAR Publishing, 2011.

Feachem, R. W. “The hill-forts of Northern Britain.” In The Iron Age in Northern Britain, edited by A.L.F. Rivet, 59-88. Edinburgh: Edinburgh University Press, 1996. Field, D. and Smith, N. Croft Ambrey, Aymestrey, Herefordshire. Investigation and analytical survey of earthworks. Research Department Report Series, 2008.

Grinsell L. V. The archaeology of Wessex. London: Methuen & Company Ltd, 1958.

Finney, J. B. Middle Iron Age Warfare of the hillfort dominated zone c. 400 BC to c. 150 BC. Bournemouth University. Doctor of Philosophy, 2005.

Güimil-Fariña, A. and Parcero-Oubiña, C. “Dotting the joins: a non-reconstructive use of Least Cost Paths to approach ancient roads. The case of the Roman roads in the NW Iberian Peninsula.” Journal of Archaeological Science 54 (2015): 31-44.

Fleming, A. “The reaves revisited.” In The archaeology of Dartmoor, perspectives from the 1990s Devon Archaeological Society Proceedings No.52 1994 (1996): 63-74.

Halliday, S. “Tap o’Noth Field visit,” RCAHMS, 1999. Accessed 20 May 2015, http://canmore.rcahms.gov.uk/ en/event/553145/.

Fleming, A. “Phenomenology and the Megaliths of Wales: a Dreaming Too Far?” Oxford Journal of Archaeology 18, no. 2 (1999): 119–125.

Halliday, S. “The Later Prehistoric landscape.” In the shadow of Bennachie: a field archaeology of Donside: Aberdeenshire, 79-114, RCAHMS. Edinburgh, The Society of Antiquaries of Scotland, 2008.

Forde, D. C., W. E. Griffiths et al. Excavations at Pen Dinas, Aberystwyth. Archaeologia Cambrensis CXII (1963): 125-153. 228

Bibliography Hamilton, S. and Manley, J. “Points of view: Prominent enclosures in 1st Millennium BC Sussex.” Sussex Archaeological Collections 135 (1997): 93-112.

Historic England. “Battlesbury Camp.” Historic England. 2015d. Accessed 19 Aug 2015. http://www.pastscape. org.uk/hob.aspx?hob_id=207399&sort=2&type=hillfo rt&rational=a&class1=None&period=None&county= None&district=None&parish=None&place=&records perpage=10&source=text&rtype=&rnumber=&p=87& move=p&nor=889&recfc=0

Hamilton, S. and Manley, J. “Hillforts, monumentality and place: a chronological and topographic review of first millennium BC hillforts of south-east England.” European Journal of Archaeology 4 (2001): 7-42.

Historic England. “Scratchbury Camp Hillfort,” Historic England. 2015e. Accessed 19 Aug 2015. http://www.pastscape.org.uk/hob.aspx?hob_id= 211396&sort=4&search=all&criteria=scratchbury&rat ional=q&recordsperpage=10

Harding, D. W. The Iron Age in northern Britain: Celts and Romans, natives and invaders. London: Routledge, 2004. Harding, D. W. Iron Age Hillforts in Britain and Beyond. Oxford: Oxford University Press, 2012.

Historic England. “Wooston Castle, “Historic England 2015f. Accessed 13 Aug 2015. http://www.pastscape. org.uk/hob.aspx?hob_id=445354

Harris, S. E. and Hounslow, M. W. Dunnideer Hillfort, Insch, Aberdeenshire: Archaeomagnetic dating of the vitrified remains. Lancaster: Lancaster University, 2010.

Hodges, R. “Knockdhu promontory fortress: An interpretation of its function in the light of some preliminary fieldwork.” Ulster Journal of Archaeology 38 (1975): 19-24.

Harris, T. M. and Lock, G. R. “The diffusion of a new technology: a perspective on the adoption of geographic information systems within UK archaeology.” In Interpreting space: GIS and archaeology, edited by K. M. S. Allen, S. W. Green and E. B. W. Zubrow, 33-53, London: Taylor and Francis, 1999.

Hogg, A. H. A. “Some applications of surface fieldwork.” In The Iron Age and its Hill-Forts, edited by D. Hill and M. Jesson, 105-126, Southampton: Southampton University Archaeological Society, 1971. Hogg, A. H. A. “Hill-forts in the coastal area of Wales.” In The Iron Age in the Irish Sea Province, edited by C. Thomas, 11-23, London: Council for British Archaeology 9, 1972.

Hawkes, C. “Hill-Forts.” Antiquity 5, no.17 (1931): 60-97. Hazell, L. C. and Brodie, G. “Applying GIS tools to define prehistoric megalith transport route corridors: Olmec megalith transport routes: a case study.” Journal of Archaeological Science 39, no.11 (2012): 3475-3479.

Hogg, A. H. A. “Excavations at Harding’s Down West fort, Gower.” Archaeologia Cambrensis 122 (1973): 55-68.

Hibbert, S. “Collections relative to Vitrified Sites.” Archaeologia Scotica 4 (1857): 280-297.

Hogg, A. H. A. Hill-Forts of Britain. Great Britain: HartDavis, MacGibbon Ltd, 1975.

Higuchi, T. The visual and spatial structures of landscapes. Tokyo: Gihodo Publishing Co. Ltd, 1983.

Hogg, A. H. A. British Hillforts: An Index. Oxford, 1979. Howey, M. C. L. “Geospatial landscape permeability modeling for archaeology: A case study of food storage in northern Michigan.” Journal of Archaeological Science 64 (2015): 88-99.

Hill, J. D. “Hill-forts and the Iron Age of Wessex.” In The Iron Age in Britain and Ireland: Recent trends, edited by T. C. Champion and J. R. Collis, 87-94, Sheffield: J. R. Collis Publications, 1996.

Hudson, E. J. “Walking and Watching: New approaches to reconstructing cultural landscapes through space syntax analysis.” In Least Cost Analysis of Social Landscapes: Archaeological Case Studies, edited by D.A. White and S.L. Surface-Evans, 97-108, Salt Lake City: The University of Utah Press, 2012.

Hingley, R. “Boundaries surrounding Iron Age and Romano-British settlements.” Scottish Archaeological Review 7 (1990): 96-103. Historic England. “Cranbrook Castle.” Historic England. 2015a. Accessed 20 July 2018. https://historicengland. org.uk/listing/the-list/list-entry/1003860.

Ingold, T. The perception of the environment: essays on livelihood, dwelling and skill. London: Routledge, 2000.

Historic England. “Prestonbury Castle.” Historic England. 2015b. Accessed 13 Aug 2015. http://www.pastscape. org.uk/hob.aspx?hob_id=445780

Ingold, T. Being Alive: Essays on movement, knowledge and description. Oxon: Routledge, 2011.

Historic England. “Cley Hill Camp.” Historic England. 2015c. Accessed 6 Dec 2015. http://www.pastscape. org.uk/hob.aspx?hob_id=207638&sort=2&type=hillfo rt&rational=a&class1=None&period=None&county= None&district=None&parish=None&place=&records perpage=10&source=text&rtype=&rnumber=&p=84& move=p&nor=889&recfc=0

James, S. and Rigby, V. “The making of the Celtic Iron Age.” In Britain and the Celtic Iron Age, 2-15, London: British Museum Press, 1997. Jerpåsen, G. B. “Application of Visual Archaeological Landscape Analysis: Some Results.” Norwegian Archaeological Review 42, no. 2 (2009): 123-145. 229

A GIS-based Analysis of Hillfort Location and Morphology Jervise, A. “Notices respecting the castle of Craig and the old kirk of Auchindoir, etc in Aberdeenshire.” Proceedings of the Society of Antiquaries of Scotland 8 (1871): 323-330.

Lock, G., Gosden, C. and Daly, P. Segsbury Camp: Excavations in 1996 and 1997 at an Iron Age hillfort on the Oxfordshire Ridgeway, Oxford: Oxford University School of Archaeology, 2005.

Kelly, L. Public and restricted ceremonial sites. In L. Kelly, Knowledge and power in prehistoric societies. Orality, memory and the transmission of culture, Cambridge: Cambridge University Press: 27, 2015.

Lock, G., Kormann, M. and Pouncett, J. “Visibility and movement: towards a GIS-based integrated approach.” In Computational approaches to the study of movement in archaeology: theory, practice and interpretation of factors and effects of long term landscape formation and transformation. Topoi – Berlin Studies of the Ancient World/Topoi – Berliner Studien der Alten Welt (23), 23-42, Berlin, De Gruyter, 2014.

King, L. J. Central place theory. USA: Sage Publications, Inc, 1984. Kӧhler, M. “Understanding the osciliating nature of hillfort settlement in Hallstat Thuringia.” In Different Iron Ages. Studies on the Iron Age in temperate Europe, edited by J. D. Hill and C. G. Cumberpatch, 163-174, Oxford: BAR Publishing, 1995.

Macdonald, J. Place names in Strathbogie. Aberdeen: D Wyllie & Son, 1891. Marsh, E. J. and Schreiber, K. “Eyes of the empire: A viewshed-based exploration of Wari site-placement decisions in the Sondondo Valley, Peru.” Journal of Archaeological Science: Reports 4 (2015): 54-64.

Kokalj, Z., Zakšek, K. and Oštir, K. “Visualisations of lidar derived relief models.” In Interpreting Archaeological Topography: 3D Data, Visualisation and Observation, edited by R. S. Opitz and D. C. Cowley, 100-113, Oxford: Oxbow Books, 2013.

McCoy, M. D., Mills, P. R. et al. “A cost surface model of volcanic glass quarrying and exchange in Hawai‘i.” Journal of Archaeological Science 38, no.10 (2011): 2547-2560.

Kormann, M. and Lock, G. “Exploring the effects of curvature and refraction in GIS-based visibility studies.” In Archaeology in the digital era: papers from the 40th Annual Conference of Computer Applications and Quantitative Methods in Archaeology (CAA), Southampton, edited by G. Earl, T. Sly, A. Chrysanth, P. Murrieta-Flores, C. Papadopolous, I. Romanowska and D. Wheatley, 26-29 March 2012, 428-437, Amsterdam: Amsterdam University Press, 2013.

McOmish, D. S. “Non-Hillfort settlement and its implications.” In From Cornwall to Caithness. Some aspects of British Field Archaeology, edited by M. Bowden, D. Mackay and P. Topping, 99-110, Oxford: BAR Publishing, 1989. McOmish, D., Field, D. and Brown, G. The field archaeology of the Salisbury Plain training area. Swindon: English Heritage, 2002.

Krist, F. and Brown, D. “GIS Modeling of Paleo-Indian Period Caribou Migrations and Viewsheds in Northeastern Lower Michigan.” Photogrammetric engineering and remote sensing 60, no.9 (1998): 1129-1137.

Mitcham, J. “In search of a defensible site: a GIS analysis of Hampshire hillforts.” In Contemporary themes in archaeological computing, edited by D. Wheatley, G. Earl and S. Poppy, 73-79, Oxford: Oxbow: 73-79, 2002.

Land Use Consultants. Wiltshire Landscape Character Assessment. London: Wiltshire County Council, 2005.

Moore, T. “Perceiving communities: exchange, landscapes and social networks in the later Iron Age of Western Britain.” Oxford Journal of Archaeology 26, no.1 (2007), 79-102.

Lee, J. and Stucky, D. “On applying viewshed analysis for determining least-cost paths on Digital Elevation Models.” International Journal of Geographical Information Science 12 (1998): 891-905.

Mytum, H. “Hillfort siting and monumentality: Castell Henllys and Geographical Information Systems.” Archaeology in Wales 36 (1996): 3-10.

Llobera, M. “Extending GIS-based visual analysis: the concept of visualscapes.” International Journal of Geographical Information Science 17 (2003): 25-48. Llobera, M., Fábrega-Álvarez, P. and Parcero-Oubiña, C. “Order in movement: a GIS approach to accessibility.” Journal of Archaeological Science 38 (2011): 843-851.

Mytum, H. C. and Webster, C.J. “A survey of the Iron Age enclosure and Chevaux-de-Frise at Carn Alw, Dyfed.” Proceedings of the Prehistoric Society 55 (1989): 263267.

Lock, G. “Hillforts, emotional metaphors, and the good life: a response to Armit.” Proceedings of the Prehistoric Society 77 (2011): 355-362.

Newman, P. The field archaeology of Dartmoor. Swindon: English Heritage, 2011. Noble, G. and Gondek, M. “Barflat excavation,” RCAHMS 2011. Accessed 27 Nov 2015, https://canmore.org.uk/ event/962065.

Lock, G. and Gosden, C. “Community and landscape- the creation of Segsbury, a new place.” In Segsbury Camp: Excavations in 1996 and 1997 at an Iron Age hillfort on the Oxfordshire Ridgeway, G. Lock, C. Gosden and P. Daly, 133-152, Oxford: Oxford University School of Archaeology, 2005.

Noble, G., Gondek, M., Campbell, E., Sveinbjarnarson, O. “Barflat Excavation,” RCAHMS 2012. Accessed 27 Nov 2015. https://canmore.org.uk/event/992702. 230

Bibliography Nolan, K. C. and Cook, R. A. “A method for MultipleCost-Surface Evaluation of a Model of Fort Ancient Interaction.” In Least Cost Analysis of Social Landscapes: Archaeological Case Studies, Salt Lake City, edited by D. A. White and S. L. Surface-Evans 67-96, The University of Utah Press, 2012.

Raftery, B. “Irish Hill-forts.” In The Iron Age in the Irish Sea Province, edited by C. Thomas, 37-58, London: Council for British Archaeology, 1972. Ralston, I. The Hill-forts of Pictland since ‘The problem of the Picts’. Rosemarkie: Groam House Museum, 2004. RCAHMS. “Tap O’Noth.” Historic Environment Scotland, n.d. Accessed 25 Aug 2015 https://canmore.org.uk/ site/17169/tap-o-noth.

O’Connor, T. “Culture and environment; mind the gap.” In Land and people, papers in memory of John G. Evans. Prehistoric society research paper 2, edited by M. J. Allen, N. Sharples & T. O’Connor, 11-18, Oxford: Oxbow Books, 2009.

RCAHMS. First Report and Inventory of Monuments and Constructions in the County of Berwick. London: HMSO, 1909.

O’Driscoll, J. “Hillforts in prehistoric Ireland: a costly display of power?” World Archaeology 1-20, 2017

RCAHMS. Sixth Report and Inventory of Monuments and Constructions in the County of Berwick. London: HMSO, 1915.

Ogburn, D. E. “Assessing the visibility of cultural objects in past landscapes.” Journal of Archaeological Science 33 (2006): 405-413.

RCAHMS. Peeblesshire: An Inventory of the Ancient Monuments. Edinburgh: HMSO, 1967.

Ordnance Survey. Wheedlemont site visit, RCAHMS 1967. Accessed 20 May 2015. http://canmore.rcahms. gov.uk/en/site/17215/details/wheedlemont/.

RCAHMS. “Dunnideer site visit.” RCAHMS 1990. Accessed 27 Nov 2015 https://canmore.org.uk/ site/1828/dunnideer.

Ordnance Survey. Land-Form PROFILE: User guide and technical specification, Southampton: Ordnance Survey, 2012. Orton, C. Sampling in archaeology. Cambridge University Press, 2000.

RCAHMS. South-east Perth: an landscape. Edinburgh: HMSO, 1994.

archaeological

RCAHMS. “Tap o’Noth-Treasured Places.” RCAHMS 2007. Accessed 25 Aug 2015. http://canmore.org.uk/ site/17169/tap-o-noth.

Cambridge:

Opitz, R. S. “An overview of airbourne and terrestrial laser scanning in archaeology.” In Interpreting Archaeological Topography: 3D Data, Visualisation and Observation, edited by R. S. Opitz and D. C. Cowley, 13-31, Oxford: Oxbow Books, 2013.

RCAHMW. “Pen y Bannau Hillfort.” Coflein n.d.a. Accessed 20 Dec 2015, http://www.coflein.gov.uk/en/ site/300704/details/PEN-Y-BANNAU+HILLFORT/. RCAHMW. “Tre-Coll, Hillfort; Trecoll hillfort.” Coflein n.d.b. Accessed 27 July 2015]. http://www.coflein.gov. uk/en/site/303565/details/TRE-COLL,+HILLFORT% 3BTRECOLL+HILLFORT/.

Opitz, R. S. and Cowley, D. C. “Interpreting archaeological topography: lasers, 3D data, observation, visualisation and applications.” In Interpreting Archaeological Topography: 3D Data, Visualisation and Observation, edited by R. S. Opitz and D. C. Cowley, 1-12, Oxford: Oxbow Books, 2013.

RCAHMW. An inventory of the ancient monuments in Glamorgan. Cardiff: HMSO, 1976.

Palmer, R. Danebury an Iron Age Hillfort in Hampshire: An aerial photographic interpretation of its environs. England: RCHM (England), 1984.

RCAHMW. “Castell Tregaron; Sunnyhill Wood Camp.” Coflein 2004. Accessed 27 July 2015. http://www. coflein.gov.uk/en/site/303561/details/CASTELL+TRE GARON%3BSUNNYHILL+WOOD+CAMP/.

Parker Pearson, M. “Economic and ideological change: Cyclical growth in the pre-state societies of Jutland.” In Ideology, power and prehistory, edited by D. Miller and C. Tilley, 69-92, Cambridge: Cambridge University Press, 1984.

RCAHMW. “Pen Dinas; Pen Dinas Elerch; Pen Dinas Bontgoch.” Coflein 2012. Accessed 27 July 2015. http://www.coflein.gov.uk/en/site/101990/details/PEN +DINAS%3B+PEN+DINAS+ELERCH%3B+PEN+D INAS+BONTGOCH/.

Parker Pearson, M. and Richards, C. “Architecture and order: spatial representation and archaeology.” In Architecture and order. Approaches to social space, edited by M. Parker Pearson and C. Richards, 38-72 London: Routledge, 1994.

Redhouse, D. I., Anderson, M., Cockerell, T., Gilmour, S., Housley, R., Malone, C. and Stoddart, S. “Power in context: the Lismore landscape project.” Antiquity 76 (2002): 945-946. Renfrew, C. Investigations in Orkney. Great Britain: Society of Antiquaries of London, 1979.

Piggott, S. “A scheme for the Scottish Iron Age.” In The Iron Age in Northern Britain, edited by A. L. F. Rivet, 1-16, Edinburgh: Edinburgh University Press, 1966.

Rivet, A. L. F. “Some of the problems of Hill-Forts.” In Problems of the Iron Age in Southern Britain, edited by S. S. Frere, 29-34, London: Institute of Archaeology, 1961.

Quinnell, H., Francis. G. et al. Spring Meeting 2013: Hillforts of Devon. The Hillfort Study Group, 2013. 231

A GIS-based Analysis of Hillfort Location and Morphology Ruestes, C. “Social organization and human space in north-eastern Iberia during the Third Century BC.” Oxford Journal of Archaeology 27 (2008): 359-386.

social interaction in the Mimbres area of southwestern New Mexico.” Journal of Archaeological Science 37, no.3 (2010): 536-548.

Sakaguchi, T., Morin, J. and Dickie, R. “Defensibility of large prehistoric sites in the Mid-Fraser region on the Canadian Plateau.” Journal of Archaeological Science 37, no.6 (2010): 1171-1185.

Ullman, E. L. “A theory of location for cities.” In The Urban Geography Reader, edited by N. R. Fyfe and J. T. Kenny 37-45, Routledge: London, 2005. Waddell, J. The Prehistoric Archaeology of Ireland. London: Wordwell, 1998.

Sanderson, D. C. W., Placido, F. and Tate, J. O. “Scottish vitrified forts: TL results from six study sites.” International Journal of Radiation Applications and Instrumentation. Part D. Nuclear Tracks and Radiation Measurements 14, no 1-2 (1988): 307–316.

Wescott, K. L. and Brandon, R. J. Practical applications of GIS for archaeologists. A predictive modelling kit. London: Taylor and Francis, 2000. Westropp, T. J. “The ancient forts of Ireland: Being a contribution towards our knowledge of their types, affinities, and structural features.” The Transactions of the Royal Irish Academy 31, 1896/1901.

Sharples, N. Social relations in later prehistory: Wessex in the first millenium BC. Oxford: Oxford University Press, 2010. Shepherd, I. Aberdeen and North east Scotland. 2nd ed. Edinburgh, HMSO, 1996.

Wheatley, D. and Gillings, M. “Vision, perception and GIS: developing enriched approaches to the study of archaeological visibility.” In Beyond the Map: Archaeology and Spatial technologies, edited by G. Lock, 1-27, Netherlands: IOS Press, 2000.

Simpson, G. “The Hill-Forts of Wales and their relation to Roman Britain: A recension.” In Dinorben. A hillfort occupied in early Iron Age and Roman times, edited by W. Gardner and H. N. Savoury, Cardiff: National Museum of Wales, 1964.

White, D. A. and Barber, S. B. “Geospatial modeling of pedestrian transportation networks: a case study from pre-Columbian Oaxaca, Mexico.” Journal of Archaeological Science 39, no.8, (2012): 2684-2696.

Stanford, S. C. “Invention, adoption and imposition.” In The Iron Age and its Hill-Forts, edited by D. Hill and M. Jesson, 41-52, Southampton, Southampton University Archaeological Society, 1971.

Wigley, A. “Rooted to the spot: the ‘smaller enclosures’ of the later first millenium BC in the central Welsh Marches.” In Later Iron Age in Britain and beyond, edited by C. Haselgrove and T. Moore, 172-189, Oxford: Oxbow Books, 2007.

Sutton, J. E. G. “Iron Age Hill-Forts and some other earthworks in Oxfordshire.” Oxoniensia (1966): 28-42. Symonds, L. A. and Ling, R. J. “Travelling Beneath Crows: Representing socio-geographical concepts of time and travel in early medieval England.” Internet Archaeology 13 (2002).

Williams, G. “Recent work on rural settlement in Later Prehistoric and Early Historic Dyfed.” The Antiquaries Journal LXVIII (1988): 30-54.

Tabik, S., Zapata, E. L. and Romero, L. F. “Simultaneous computation of total viewshed on large high resolution grids.” International Journal of Geographical Information Science 27 (2013): 804-814.

Wiltshire Council. Extract from Wiltshire and Swindon Historic Environment Record. Chippenham: Wiltshire Council, 2013. Wood, J. Landserf. J. Wood 2009. Accessed 13 November 2015. www.landserf.org.

Taliaferro, M. S., Schriever, B. A. and Shackley, M. S. “Obsidian procurement, least cost path analysis, and

232