Little Solsbury Hill Camp Geophysical Survey Batheaston, Somerset 2012 John Oswin and Rick Buettner Bath and Camerton Archaeological Society Forward by Mark Corney Coauthor Wessex Hillforts Project Geology by Vince Simmonds Figure I Weaving combs from W.A. Dowden excavations on Little Solsbury Courtesy University of Bristol Speleological Society Report compiled by Rick Buettner and Jude Harris © Bath and Camerton Archaeological Society 2014 Geology photos by Vince Simmonds All other photos by Rick Buettner unless otherwise acknowledged The survey was done under English Heritage License 22677 Forward Hillforts are the dominant monuments of the later prehistoric period still visible in the landscape and occur across a wide swathe of Britain. There are marked concentrations of large hillforts on the Wessex chalk, the Welsh Marches and the Cotswolds. Little Solsbury Hill , just to the east of Bath, is the southernmost example of the Cotswold group and relatively small scale excavations at between 1955 and 1958 provided an outline sequence but provided little detail of the interior or structural evidence for occupation. Since those trenches were backilled there has been no further work on the site Hillfort studies have been, and to some extent still are, heavily biased towards the chalklands of Wessex. This began with General Pitt-Rivers excavation of Winkelbury on the edge of Cranborne Chase in the late nineteenth century and continued between the two World Wars with a series of major investigations by Christopher Hawkes in Hampshire and Sir Mortimer Wheeler at Maiden Castle in Dorset. The interpretation of those investigations were to have a profound impact on Iron Age studies for a generation and remained unchallenged until the widespread application of radio-carbon dating methods began in the 1960s. Since this period analyses of artefacts, landscapes and environmental data have revealed a far more complex situation demonstrating the origins of hillforts to be complex with origins in the late Bronze Age. In recent decades the focus on Wessex hillforts has continued, with further major excavations at Maiden Castle, South Cadbury and Balksbury near Andover. The most inluential project however remains the major campaign of excavation and accompanying hinterland surveys centred upon Danebury under the direction of Professor Sir Barry Cunliffe. The cost of large scale research excavation on the Danebury model is enormous and the advent of non-intrusive survey methods now provides a cost effective means of acquiring detailed plans, often covering large areas, where the local soil and geological conditions are favourable. Geophysics has now become a well-established investigative tool for archaeology. The method is ideal for application over large sites such as hillforts. Even in this ield the dominance of Wessex has continued with the South Cadbury Environs Project and the Wessex Hillfort Project. The latter was undertaken by English Heritage, published in 2006, and produced a series of detailed surveys demonstrating the diversity of internal morphology, and by implication chronology and function, within a single region. The investigation of Cotswold hillforts has been poor in comparison to neighbouring Wessex although the region boasts some large and spectacularly located examples. Apart from Crickley Hill near Birdlip Edge and Danes Camp, Conderton, no large scale excavations have been undertaken and the application of geophysics is commensurately limited. The advent of relatively affordable sophisticated geophysics equipment has revolutionised the application of the method allowing wider access to the technology. The Bath and Camerton Archaeological Society has an established reputation for producing high-quality non-intrusive surveys and reports and it must be regarded as one of the leading regional societies regularly applying geophysical survey to a wide range of sites and landscapes. The results of the survey of Little Solsbury are a major contribution to the study of Cotswold hillforts. It has revealed clear evidence for permanent settlement in the form of roundhouses, these being clearly visible despite the overlying earthworks created by later medieval cultivation. Of especial interest is the discovery of a hitherto unknown large ditch cutting off the north eastern apex of the hillfort circuit. This has a single entrance with slightly offset terminals and creates a small enclosure within which possible post-hole clusters may be discerned. This intriguing feature would be an ideal target for small scale excavation to ascertain function and date. The authors of this report and their dedicated supporting team of volunteer ield workers are to be heartily congratulated on the success of the survey and the new data it has provided. I believe iii this project and report to be a model for what local societies can achieve and it provides a secure base for any future investigations at Little Solsbury Hill. Having surveyed Little Solsbury Hill with such thoroughness I have to ask the authors whether this could not be the beginning of a more ambitious long term project to examine other Cotswold hillforts and provide regional data sets to compare with those from neighbouring Wessex? Mark Corney iv Table of Contents Forward iii Contents v List of Figures vi Acknowledgements viii 1 Introduction and Geology 1 2 Previous Work 3 3 Geophysical Method 9 4 Geophysics Results 17 5 The Hillfort in its Landscape 35 6 Comments and Conclusions 40 Bibliography 43 Appendix A Details of Gridding and Surveying 44 Appendix B LIDAR 47 Appendix C Views of Little Solsbury 48 Appendix D Batheaston Freeholders Guidelines 53 Appendix E Details of Grid Location 53 v List of Figures I Weaving combs from Little Solsbury excavations ii 1-1 1-2 Exposed Oolite layers. Oolite slab with striations. 2 2 2-1 2-2 2-3 2-4 2-5 2-6 2-7 Dowden excavation 1958 Field terraces today Photomap of Little Solsbury location Map of Little Solsbury from Dowden excavation 1840 Tithe map of Solsbury Solsbury from the northeast aerial photo LIDAR image of Little Solsbury hilltop 4 4 5 5 6 7 7 3-1 3-2 3-3 3-4 3-5 3-6 3-7 3-8 Bartington 601-2 dual luxgate gradiometer on Solsbury Hill Bartington MS2 magnetic susceptibility meter on Solsbury Hill TR/CIA twin probe resistance meter on hilltop overlooking A46 RM15-D twin probe resistance meter on hilltop overlooking Bath Location of GPR and 1m probe grid RM15-D twin probe resistance meter 1m separation on Solsbury MALA X3M ground-penetrating radar Leica Smartworx Viva GPS survey on Solsbury Hill 10 10 11 12 13 13 14 15 4-1 Magnetometry results 4-2 Exposed section in 1931 quarry from Falconer, J.P.E., and Adams, S.B. 4-3 Inner ditch on the southern end of the hill. 4-4 Annotated magnetometry plot 4-5 Magnetic susceptibility plot 4-6 Twin probe resistance plot with dark high values 4-7 Twin probe resistance plot with pale high values 4-8 Resistance and magnetometry overlay 4-9 Northern rampart 4-10 Annotated twin probe resistance 4-11 Inner ditch with 1 metre probe separation 4-12 .5 metre and 1 metre probe separation comparison 4-13 Double boundary stone in NE cordon ditch 4-14 GPR inner ditch depth slices 4-15 GPR NE cordon ditch depth slices 4-16 Precision GPS contour survey 1 m intervals 4-17 Southern end of hillfort .2m intervals 4-18 Central hilltop .2m intervals 4-19 Northeast corner .2m intervals 4-20 Northwest corner .2m intervals 4-21 Southeast entrance .2m intervals 4-22 Possible occupation features 4-23 Decorated Little Solsbury pottery sherd 17 18 18 19 20 21 22 22 23 24 25 25 25 26 27 28 28 29 30 31 32 33 34 5-1 5-2 35 36 Solsbury Hill in the local landscape Solsbury viewshed vi 5-3 Solsbury spearpoints Panoramas from the hilltop 37 38 6-1 6-2 6-3 Rampart trench Dowden 1958 Batheaston Entrance Little Solsbury from hot air balloon over Lansdown 40 41 42 A-1 A-2 A-3 A-4 Magnetometry grids Twin probe resistance grids Inner ditch grids Northeast cordon ditch grids 44 45 46 46 C Views of Little Solsbury from surrounding locations 48 D-1 Batheaston Freeholders Rules for Quarrying 53 E 54 Grid location photo aids vii Acknowledgments I undertook this project to satisfy my curiosity about Solsbury Hill, which is an unavoidable ixture in the local landscape in Batheaston where I live. With just enough background to know a little about archaeology, it was apparent to me that the hill must have many hidden secrets. As a member of BACAS (The Bath and Camerton Archaeological Society), I knew where to look for help and support in this endeavour. It turned out to be a large undertaking involving BACAS members, national organisations, and the local community. We would like to thank Hugh Beamish and English Heritage for help obtaining the Section 42 license which was necessary before starting the survey. Martin Papworth of the National Trust was also very helpful in the permissions stage and encouraging throughout the project. Michael Tabb and the Batheaston Freeholders were very supportive of the project and helped with scheduling the cattle grazing on the hilltop. Steve King, the owner of the cattle, was very helpful with bovine matters. Richard Sermon, BANES Archaeologist, also supplied excellent advice. Robin Holley, Ceri Lambdin, and Bob Whitaker at BACAS all helped with the necessary paperwork, and Bob supplied most of the historical references. John Samways helped get the work going with his advice and enthusiasm. Chris Ellis kindly made a contour survey of the entire hilltop. Vince Simmonds supplied the geological expertise, and Tim Lunt contributed the viewshed and landscape information. Janet Pryke helped plot the magnetic susceptibility points. Jude Harris has helped put it all together for publication. There were many BACAS members pressed into service on the hilltop. Surveying over 20 acres involved many days work. In no particular order, Roger Wilkes, Jan Dando, Janet Pryke, John Knapper, Ian Perkins, Les Phillips, Sharn Pepperell, Bernadette and Oliver Stanley, Chris Higgs, Sue Rhodes, Debbie Watkins, Sandra Heward, Pauline and Robert Cheyne, Andy Kingan, Wendy Russ, John Richards, Carole Bishop, Sue Pickering, Nicola Carroll, Richard Wyatt, Vince Simmonds, Tim Lunt, Laurence Chadd, John Branston, Rod and Liz Thomas, Peter Martin, Lawrie Scott, Sophie Hawke, Julie Bassett, Patrick McCarron, Paul Henaghan, Joanne and Isobel Rippin, David Hughes, Mary Huntley, Gillian Vickery, Laurie Bingle, Charles Singer, and Sarah Price all spent time up on the hill in all weathers, not that the hilltop isn’t a splendid place to be. My thanks again to all of you. Finally, none of this would have happened without John Oswin. John’s expertise in geophysics and general knowledge of running projects allowed me, as a novice in these areas, to organise what I hope is a valuable contribution to the archaeology of Solsbury Hill. Rick Buettner viii 1 Introduction 1.1 Location Little Solsbury Hill stands in a dominant position overlooking the River Avon, the Fosse Way and the A4 northeast of Bath. It lies in the Parish of Batheaston in what is now Bath and Northeast Somerset just east of the A46 Bath to Gloucester Road. It can be considered part of the southern end of the Cotswolds. The higher hill of Charmy Down is across a small valley to the north, Lansdown is to the west across the A46 and Woolley Valley. The Avon Valley sweeps up in front of the hill from the southeast, turning to the southwest and Bath. Across this valley is Bathampton Down. Directly east of the hill lies the St. Catherine valley and Batheaston. Above Batheaston on that side, Bannerdown is another higher prominence. The Bybrook Valley separates Bannerdown from Bathford and Kingsdown on the east of the Avon Valley. There are ine views from Solsbury of most of the country around Bath, all the way to Devizies in the east, Westbury and Bratton in the south, and to King Alfred’s Tower and the Mendips in the southwest. Standing as an isolated hill with a latish top, steep approaches on all sides and open views of most of the surrounding country, Solsbury Hill was an excellent location for a settlement. Well into historic times Little Solsbury has been used and exploited for agriculture and stone by the adjoining communities of Batheaston, Swainswick and Larkhall. Today the hilltop is owned by the National Trust and is protected from overuse except perhaps by dog walkers. 1.2 Geology Little Solsbury Hill: A Geological Overview by Vince Simmonds The British Geological Survey 1: 50 000 scale map describes the bedrock geology that underlies Little Solsbury Hill as Lower and Middle Jurassic (undifferentiated) – interbedded limestone and mudstone. These rocks comprise mainly siliciclastic sediments that have been deposited as mud, silt, sand and gravel in shallow sea, about 159 to 206 million years ago (BGS online, 27/01/2013). Bath is the type area for the Great Oolite (Middle Jurassic) and it is this group of rocks that form the scarps and plateaux that are so distinctive of the area (Green, 1992: P134). At Little Solsbury Hill it is the prominent rock type forming the summit and outcrops can be observed in the disused quarries along the western and northern slopes. The Great Oolite succession has been described by Green and Donovan (1969) and is summarised in the table below: Thickness (m) [Forest Marble, Upper Rags] Bath Oolite - 5 – 8.5 Twinhoe Beds 2.5 – 8 Combe Down Oolite 10 – 17.5 The Oolite types (Bath and Combe Down) can generally be described as cream coloured, crossbedded oolite freestone; the top surface is usually planed and bored, with the base generally sharp and occasionally erosional. There are good exposures of the rock type in the quarries previously mentioned above and an example is pictured below (Figure 1-1). On the plateau above the quarry there are two isolated blocks of oolitic limestone, the slab of rock located at NGR ST 76714/68016 ± 6m has a number of interesting striations. The rock lies approximately 2 metres west of a lynchet (Figure 1-2). The second block located at NGR ST 76751/68025 ± 6m appears to be situated at the north end of an area of very slightly raised ground. 1 Figure 1-1 Exposed Oolite layers Figure 1-2 Oolite slab with striations There is a spring line along the 170 metre contour on the northern and eastern slopes and the active springs indicate there is a perched water table. There is some evidence of a landslip on the northern slope where a large area of bedrock appears to have slumped. The eastern and southern slopes also have some signs of landslip, probably the consequence of the small scale quarrying in the form of stone pits that has occurred in the past. 1.3 Team and Times The top of Solsbury Hill appears lat at irst glance, but it is actually dome shaped and around 20 acres in area. We planned to use the magnetometer over the entire hilltop and to do as much resistance measuring as possible. This necessitated a large number of volunteers working over some months. The exposed position of the hilltop and the presence of cattle over the summer further complicated scheduling. The brunt of the work including all the magnetometry was done in April before the cattle were introduced. We returned in September and inished most of the resistance and the mag sus. Since we were close to inishing the resistance, we applied for an extension to the Section 42 License, due to end on September 30th, which we received. This enabled the team to return in October to inish the resistance and use the ground penetrating radar and the 1 metre resistance bar in a few areas. We inished on Halloween, having lost a few days to rain and suffering a good deal of cattle interference in September and October. 1.4 Scope This report covers the geophysical survey work on Solsbury Hill completed in October 2012 and adds it to previous surveys, descriptions and excavations to create an up to date compilation of the information known about Solsbury Hill. Some interpretation comes from new general sources of Iron Age hillfort information such as the Wessex Hillforts Project sponsored by English Heritage. 2 2 Previous Work 2.1 Written sources Solsbury Hill is irst mentioned by Collinson (Collinson 1791, 98-101) as Salisbury Hill. He says the Antiquaries claim there was a temple to Bladud on the top and the visible remains were a Saxon fort used in the siege of Bath in 577 AD. At that time barley was grown on the summit. A more prosaic assessment of the hill was made by Phelps (Phelps, 1836, 102-102) where he describes the features and geology of the “Sulisbury” hilltop in detail without any historical attributions. These details are essentially the same as seen today. John Evans visited the hill, which he referred to as Little Solsbury, in 1862. Whilst ield walking, he found much evidence of lint work, which would seem to mostly date from the Neolithic, and pottery that he placed in the Iron Age (Evans 1866, 240-243). Collins and Cantrell visited “Solisbury, Sulisbury, or Little Solsbury as the natives prefer to call it” over an eight year period (Collins and Cantrell 1909). They were able to achieve a relatively thorough understanding of the site by observation and studying of the quarry sides and other inroads into the fabric of the hilltop. They found a burial in the side of one quarry and much pottery, many animal bones, and evidence of metal working in the same general area, around the northwest corner of the hilltop. Noting Evans’ earlier work, they were surprised to ind only a few stone or lint items. They concluded the site had been irst occupied in the Neolithic, and the main part of the occupation dated from the bronze age through the early Iron Age. The irst modern scientiic analysis of items from Little Solsbury was made by Falconer and Adams (Falconer, Adams 1935, 183-221). They revisited the same area where Collins and Cantrill had found most of their material, adding more inds of a similar nature. They analysed their own new material and that of Collins and Cantrill, which was stored in Bath. In-depth comparisons of the pottery inds with the established pottery style time lines led them to conclude the hilltop was occupied from the early Iron Age to the middle Iron Age and then abandoned. In 1955 there was an actual excavation on Little Solsbury (Figure 2-1). W. A. Dowden dug three series of small trenches in 1955, 1956, and 1958. (UBSS, Dowden 1957, 1962). These excavations were on the northwest hilltop in the vicinity of the quarry visited by Collins and Falconer. Taken down to bedrock, the trenches showed three phases of occupation. These were pre-rampart, rampart era, and post-rampart. The rampart had been constructed some time after the irst Iron Age occupation began and had apparently been thrown down for an unknown reason. Building and occupation continued well after this event. The inds were broadly similar to those of Collins and Falconer. The bedrock surface had been levelled where uneven and had several postholes. These postholes had apparent ritual pottery deposits in the bottom and showed no evidence of burning, leading Dowden to suggest there had not been a violent event with widespread burning. Dowden also concluded that most of the occupation had probably been along the inside of the rampart as he found no evidence of huts in his innermost trenches, some 40 feet from the rampart (UBSS, Dowden 1957, p.23). His inds and analysis supported the dating time line put forward by Falconer. 3 Figure 2-1 Dowden excavation 1958 Today the medieval and later ield systems are the most prominent features on the hilltop after the remains of the rampart. Beatrice Wilmot Dobbie delves into the history of Solsbury Hill as common land and includes a reproduction of the 1840 tithe map which shows the terraces and the farmers responsible for the various parts and the marker stones delineating these divisions (Figure 2-2). (Dobbie 1969: 134-143). Figure 2-2 Field terraces today 4 2.2 Maps and drawings Figure 2-3 Photomap courtesy Google Earth Figure 2-4 Solsbury map from Dowden UB 5 Figure 2-5 1840 tithe map of Solsbury Hill showing terrace system ref. D/D/rt/M/352 Somerset Archives and Local Study Service 2.3 Aerial photographs and LIDAR The aerial view of Little Solsbury (Figure 2-6) illustrates the advantages of the hilltop for defensive purposes. The location of the lat plateau with sudden steep slopes on all sides in an isolated situation with further long slopes down to the valley loors makes access from any direction a long uphill approach with a formidable natural barrier at the top. The addition of a rampart behind a 12 foot high stone wall would have made any hostile incursion dificult at the least. Whether defences were for display or actual deterrence or both is up for question with Iron Age hillforts. This view also shows the medieval terracing, the most prominent features on the top today. These were in use well into the 19th century, and cultivation of barley continued until the 1950s. The top is not actually lat, but rather dome-shaped with the highest point near the centre on the northwest side. The terracing follows the contours of the hill along the sides, which would have helped retain the topsoil and rainwater. Small boundary stones mark the ends of a farmer’s section of the terrace. It is also possible to note the effects of quarrying and slumping in this photo. The large chunks taken out of the right hand end of the hilltop were quarries for lining the reservoirs built in the valley bottom near Northend in Batheaston. The rampart that surrounded the hilltop is almost entirely gone, which indicates a large amount of opportunistic quarrying by locals and some geological slumping as shown by the lumpy deposits at the bottom of the hilltop edge. 6 Figure 2-6 Solsbury Hill from the northeast Copyright reserved Cambridge University Collection of Aerial Photography Figure 2-7 LIDAR of Solsbury Hill © Environment Agency copyright 2013. All rights reserved 7 These effects are also shown well in the 1 metre LIDAR image (Figure 2-7), kindly supplied by Geomatics, part of the Environment Agency. This technique uses multiple lasers to see through vegetation and structures (See Appendix B). In this case, the terraces become very prominent, and there are several visible that cannot be seen on the ground. The two quarries stand out on the top left corner, and both entrances register prominently. Waves of slumping can be seen, particularly on the right side and bottom. On the left side, there is debris from small scale quarrying of the rampart and hillside. At the very bottom of the plateau, it is possible to pick out the only visible section of the inner ditch. There are several well-worn pathways leading away from the hilltop around edges and entrances. 8 3 Geophysical method 3.1 Gridding With an area of over 20 acres to survey, organisation and eficiency in the gridding process were essential. A lack of permanent reference points for relocating the grids at a later date was also a problem. The lat area of the plateau to be surveyed forms a triangle with the widest expanse at the north end. This end of the hill faces almost due north. Bearing all this in mind, we decided to start along the north end of the hill with the irst full row that would cross the hilltop from the western edge to the eastern edge. For the initial two grid points we located the northern extreme of the complete row of grids at the point where the rampart gave way to the steep slope down. We surveyed these points in using handheld GPS and compass so that the grid pattern could be overlaid on maps, but these grid references were only accurate to 5 m, so were not suficient for relocating grid posts. Starting the grid here, we left markers in the ground for the two reference points. All grid points were measured in using 100m tapes. Before leaving the hilltop at the end of April, these points were measured in using more permanent stones for reference. The gridding continued south from this row quite quickly to accommodate the magnetometry work. Markers were left along the top two rows and around the edges for the resistance, but the markers in the middle were removed as the work progressed. There was some interference with markers by members of the public, who thought, among other things, that we were laying out a golf course. Upon reaching the southern end of the hill, we had marked out over 250 grids. At the end of the irst phase in April, we measured in grid points with permanent references on the east, west, and south sides of the hill in addition to those on the north. At this point we had inished the magnetometry, but the resistance had only covered the northern two rows and most of two or three squares in from the edges around the sides of the hill. This project was lexible in scope, depending on the results as we proceeded to indicate how much time and further work BACAS would attempt on Solsbury. In the event, the magnetometry was spectacular, and we decided to return for more work in the Autumn. Unfortunately, this meant taking up most of our grid markers and starting over in September. We were able to leave some stakes and more permanent markers, and enough of these survived to reconstruct the grid system reliably. The permanent reference points were used to check the grid, and we were successful in keeping the grid matched with the earlier work. We were able to take reliable measurements for triangulation from both the trig point and from the memorial stone, and as these were both towards the southern side of the plateau, we could use these for reconstructing the grid later in the year. Boundary stones and the large lat stones with striations mentioned in the geology section were used for the other points. The triangulation data are given in Appendix A. 3.2 Magnetometry The magnetometer used was a Bartington 601-2 dual luxgate gradiometer, as shown in Figure 3-1. The device was set to take readings at 0.25 m intervals along traverses 1 m apart, giving 1600 data points per 20 m grid square. Grids were started at the south-west corner, traverses heading northsouth. Grids were started 1 m from the corner, 1 marker north of the south base line, in accordance with Oswin, 2009, igure 5.8c page 115. Approximately 250 grids (some partial) were surveyed. The survey was completed rapidly 9 Figure 3-1 Bartington 601-2 dual luxgate gradiometer on Solsbury Hill 3.3 Magnetic Susceptibility A full area random point magnetic susceptibility was undertaken using a Bartington MS2, itted with 200 mm coil. This is an old device with no data logger, so locations of readings were logged on a hand-held GPS. This was precise to 1 m but only accurate to 5 m. The readings and locations were written down and then transcribed into a spreadsheet and a graph plotted using DPlot freeware. This also meant that reading points could not be located instantly on the grid used for magnetometry but the area covered extended off the edge of the fort, beyond the ramparts, beyond the survey grid. However, the magnetometer plot could be overlaid with reasonable accuracy on the main survey grid afterwards, for comparison with those results. Approximately 1000 points were logged. The magnetic susceptibility meter is illustrated in Figure 3-2 Figure 3-2 Bartington MS2 magnetic susceptibility meter on the hilltop . 10 3.4 Twin Probe Resistance Two twin-probe resistance meters were used, but always kept a good 60 m apart to avoid any crosstalk between them. One was a TR/CIA, as illustrated in Figure 3-3 and the a Geoscan RM15D, shown in Figure 3-4. Both took readings at 0.5 m intervals along traverses 1 m apart, using the same grids as the magnetometer, and used on the same basis. Figure 3-3 TR/CIA twin probe resistance meter These could have caused problems, but use of INSITE processing software allowed mixing of the two meters’ outputs, as one downloaded series data and the other downloaded zig-zag data. 11 Figure 3-4 RM15-D twin probe resistance meter Only about 100 grid squares were complete by the end of April, the initial survey period. The survey was resumed in September and continued until the end of October. The initial survey had tried to complete the periphery, but this was not completed then. From September, the area was illed in. Early September saw one of the few dry periods in the year, and the grids surveyed in this period had different degrees of contrast so they did not match the earlier and later grid squares. A few grids were surveyed using the RM15 device itted with one-metre probe separation in order to see to greater depth, particularly over a section over the inner ditch along the north rampart and also over the ditch which separates off the northeast enclave. See Figures 3-5 and 3-6. Ideally, the TR/CIA device would also have been used in its electrical pseudo-section mode, but this was not possible for safety reasons, both in terms of carrying the very heavy cable looms up the steep slopes and also deploying the looms in an area used by cattle, dogs and the public. 12 Figure 3-5 Location of GPR and 1m probe grid Figure 3-6 RM15-D twin probe resistance meter 1m separation on Solsbury 13 3.5 Ground-penetrating Radar The radar used was a MALA X3M. It was used on one site with a 500 MHz head to try to obtain more details expected to be shallow. It was also used on two areas using a 250 MHz head (see Figure 3-7). Area 1 (500 MHz) was 40 m square, as was area 3. Area 2 was 18 m north-south and 20 m eastwest. All were surveyed in zig-zag mode along north-south transverses 1 m apart. The areas were based on existing grid squares used for resistance measurement. Area 2 was truncated at the northern end by the steep rise of the rampart. Areas 1 and 3 were equivalent to a block of four grids. The 500 MHz head collected data every 0.05 m, the 250 MHz every 0.1 m. The areas subject to radar and to resistance survey with 1 m probe separation are shown in Figure 3-5. Figure 3-7 MALA X3M ground-penetrating radar 3.6 Contour mapping We are very grateful to Chris Ellis and to Wessex archaeology for making this survey possible. The intention was to produce a very detailed contour survey, suficient for contours of 1 m. A number of ixed points were surveyed in (see section 3.1) but most points were measured by walking long traverses over the hill top. Areas outside the ramparts were also measured so that the steep slopes could be analysed in detail. Results were plotted in Dplot. Surveying is shown in Figure 3-8. The device was a Leica Smartworx Viva. 14 Figure 3-8 Leica Smartworx Viva GPS survey on Solsbury Hill 3.7 Software The principal data processing software was INSITE v4. This is obsolete, but is still the processing package preferred by BACAS as it gives a very visual approach and is very versatile in gridding. It is now supplemented by BACAS proprietary download software for TR/CIA, Geoplot RM15D and FM256. The Bartington magnetometer downloads via its own software. INSITE has also been supplemented by BACAS proprietary zero-median de-stripe software for magnetometers. Some of the data were also analysed and displayed using a spreadsheet routine in XL. This allowed multicolour plotting and linear resistance scale to give iner detail. Where data points were not on a rectangular grid pattern, for instance with magnetic susceptibility, data collected in XL were displayed in DPlot freeware. Radar data were analysed and displayed in REFLEXW software. GIS displays were developed in Global Mapper freeware. 3.8 Constraints The extent to which partial grids round the edge of the survey was determined by the steep slopes, with equipment operators going only as far as they assessed it to be safe. Although the magnetometer survey was completed within three weeks, the resistance survey had to be completed in three separate time spans: April; September; late October. This gave some variation in ground conditions, although the very wet year meant good matching, apart from some drying out of the soil in early September. Magnetometer anomalies were strong. The biggest problem was inding a location quiet enough for calibration. The post-mediaeval plough lines tended to dominate the print-out, but earlier features could be seen below. The weather was generally benign; indeed the hilltop seemed to be sheltered from the worst of the storms that sometimes aflicted the area. Heavy rain curtailed operations only on two or three days. 15 This was an entirely unfunded operation, performed by volunteers from BACAS, using their equipment. The only exception was contour survey volunteered by Chris Ellis of Wessex Archaeology, using their specialist equipment. In general, the best view of the results is seen on the computer screen. Some detail can be lost when the screen contents are printed to a document, particularly from a low resolution screen document. 16 4 Geophysics results 4.1 Magnetometry The plot of magnetometry is shown in Figure 4-1. It is instantly obvious that the plot is very lively. Indeed, the biggest problem was not inding anomalies, but inding a suficiently quiet location for calibration. Signals were typically of the order of 10 nT, so noise was not a problem. The whole site also seemed remarkably free from ferrous anomalies. There were particular sites of ferrous signal, but these were at the trig point and at the memorial, where iron ittings might be expected. The most obvious features are the striations which run across lengths of the monument at various angles. These relate to the mediaeval and post-mediaeval open ield strips which occupied the hilltop, and which remained in use during the nineteenth century. The major lynchets are generally visible on the surface of the hilltop; some are even stone-built, but there are also plough lines within these. Fortunately, other features can be seen beneath these plough lines, but there are places where they can appear to give extra detail, and so add confusion. Figure 4-1 Magnetometry results Below the mediaeval features, a dark feature around the whole periphery of the hillfort represents a ditch inside the ramparts. Along the northern edge, the a portion of the ramparts are evident as ‘neutral’ beyond the strong positive signal of the ditch. Along the eastern and western edges, the ditch follows the edge of the plot, where the ramparts are missing, either by slumping or from quarrying. A section in a quarry near the northwest corner in 1931 (Falkner and Adams 1935) showed the ditch to be full of occupation material. See Figure 4-2. Excavations in the ditch nearby (Dowden 1957: 1962) reported signs of burning. Some of this material may have been deposited in the ditch, other material may have been washed in as soil has moved down the slope from the interior. The ditch is not visible on the surface other than at the extreme south of the hillfort, where it remains about 1 m deep over a stretch of some 60 m (Figure 4-3). 17 Figure 4-2 Exposed section in 1931 quarry from Falconer, J.P.E., and Adams, S.B., UBSS, 1935. Figure 4-3 Inner ditch on the southern end of the hill. Another ditch cuts off a northeastern enclave, although there are gaps at either end and in the centre. This is indicated as [1] in Figure 4-4. The area within this enclave seems to differ in its magnetic responses from the main area of hillfort. A large area of anomaly at the far north-east may represent a dew pond [2], as it is also the lowest point on the hilltop, with ramparts beyond to act as bunds. Generally within the northeast enclave, anomalies seem to represent groups of post-holes, possibly representing raised structures. Perhaps this area was set aside for storing provisions. The north-west entrance seems to have a pair of ditch-like anomalies [3] funnelling out from it for some tens of metres, presumably part of the entrance control, and there appears to be a row of post holes just to the south of the entrance, inside the line of the inner ditch [4]. Similar but less widely spaced lines funnel out from the southeast entrance [5] and this case are directed at a prominent round house [6] at the middle of the hillfort. This is curious, as the southeast entrance has been considered modern. The approach to it is certainly very steep, making for very dificult access. It is 18 possible that the northern of these two lines is connected with the mediaeval ield system as it seems to terminate some of the lynchets, but it is more likely to have pre-dated these features. There is also a prominent very large elliptical feature further to the south [7], but given its dimensions, some 25 m by 20 m, it may possibly represent a barrow. Given its position, it would probably appear on the false crest of the hill from the south. If it is a barrow, it seems to be a lone monument and it is not visible on the surface at all. Indeed, there appears to be a slight depression here. This feature appears to have internal structure, but this is almost certainly an overlay of mediaeval plough features. Apart from these prominent circular features, the whole of the hilltop (apart from the north-east enclave) is covered with signatures of ‘round houses’, numbering some 20 or 30. Occasionally one may appear superimposed on another, [8] for instance, suggesting some replacement. There may be other features, some pit-like, associated with these, but it is not possible to distinguish these as being deinitely Iron Age, rather than overlay of mediaeval ploughing features. Figure 4-4 Annotated magnetometry plot 4.2 Magnetic Susceptibility The plot of magnetic susceptibility (*105 SI units) is shown in Figure 4-5. The scale has been set arbitrarily in steps of 10. Readings were up to 750 in high spots, but recording very high values under separate colours would add no beneit. The scale chosen gives good discrimination of areas of various levels. Note that a closed plot is formed, so values shown around the periphery may not be entirely valid. The hand-held GPS used was only accurate to 5 m although it gave readings to the nearest metre. The resolution of the plot can therefore be considered to be only 5 m at best, but this is suficient 19 to show areas of signiicant variance of levels. Use of GPS automatically provides national grid localisation of areas but is not instantly compatible with the magnetometry and the resistance plotted in local coordinates. A number of features stand out from the magnetic susceptibility plot and these complement the magnetometry shown in the previous section. The generally high levels within the hillfort are compatible with the very lively magnetometer plot, again suggesting that all the effects are contained within a shallow surface layer, as the penetration of the MS2 does not exceed 200 mm. Levels are generally much lower outside the hillfort, particularly along the western and eastern sides where the ramparts have been obliterated. One might expect washed-out ditch ill to cover the ground and give moderately high readings, but perhaps centuries of precipitation has washed this material on down the slope and diluted it further. The most obvious feature is the string of very high readings representing the inner ditch, and corresponding with the magnetometer output. The ditch is assumed to be full of wash off the higher interior of the hillfort, and this is rich in occupation material for the human occupation areas. It is noticeable that high levels in the ditch do not extend into the northeast enclave, although this is the lowest area in the hillfort. Either all the high susceptibility material has been trapped in the ditch which cordons it off, or there was much less human occupation of this area. The second option is the more likely as the cordon ditch itself does not show as a high susceptibility feature. A section was cut in a quarry face in the northwest corner (where there are very high readings) in 1931, although it was damaged before it could be properly recorded (Falconer and Adams 1935) and a photograph of this was reproduced in Figure 4-2. This appears to show dark ditch material above the lighter coloured rock. Many inds were recorded in this section, suggesting an accumulation from the hilltop settlement, and excavations in the 1950s (Dowden 1957, 1962) in the ditch close by suggested occupation debris and burning, all likely to enhance the magnetic susceptibility. Figure 4-5 Magnetic susceptibility plot. The area extends beyond the ramparts. 20 4.3 Twin Probe Resistance Two versions of the resistance plot are given: Figure 4-6 shows the greyscale plot where dark represents high values; Figure 4-7 shows the same plot reversed, with pale representing high values. The author prefers the version of Figure 4-6 and considers this complements the magnetometry best, but the reversed version is used regularly by other operators, and it does pick out a few features which are less obvious in Figure 4-6. Edge matching is not perfect across the plot. It is more important that detail within any grid square is clearly visible. Also, the one period of dry weather in early September did cause the soil to dry out, giving a rather more speckled, high contrast effect. This may have lost some detail compared with the moister conditions. Figure 4-6 Twin probe resistance with dark higher values A red-shaded version of the plot has also been produced, and this was used in a ‘dark is high reading’ mode to overlay on the magnetometry, which was green-shaded. This overlay is shown in Figure 4-8. 21 Figure 4-7 Twin probe resistance with pale higher values Figure 4-8 Resistance and Magnetometry overlay 22 An annotated version of Figure 4-7 is shown in Figure 4-10, and numbers in the following paragraphs pertain to this igure. The inner ditch is visible here as a pale patch running round the monument. Beyond this, at the north [1], the rampart where it survives (Figure 4-9), shows as a very stony feature. Many of the mediaeval and post-mediaeval lynchets are picked out, normally as pale, but plough lines are less evident, so that archaeology beneath them can be picked out better. The lower resistance lines here represent build up of soil against the inside of the lynchet revetment. The northeast cordon ditch is visible as a pale feature [2]. This ditch will be discussed in more detail later in this section. Figure 4-9 Northern rampart A number of the ring ditches are visible as pale features across the hilltop, and this includes the large prominent hilltop roundhouse [3], corresponding to Figure 4-4[6]. The very large elliptical feature (Figure 4-4[7]) cannot be picked out in resistance, suggesting a different nature. If visible it would be at point [4], partly on high resistance, partly on low. The funnelling ditches at the northwest [5] and southeast [6] entrances are visible as low resistance features. In the far northeast corner, there appears to be a partial ring [7] where a magnetic feature was interpreted as a possible dew pond (Figure 4-4[2]). A large area of high resistance just within the northeast cordon ditch [8] appears to have holes cut in to it, which may be the equivalent of the post holes showing in magnetometry. There appears to be a stony feature set within low resistance towards the north [9]. It is about 4 m across. Its purpose is unknown. There is a scatter of low resistance ‘holes’ typically 3 m across, over the hillfort. These might be pits. They may show on magnetometry but they are obscured by the plough overlay. Just south of the centre, there appears to be a set of rock-cut features some 10 m square [10]. The eastern line of this may be part of a longitudinal feature, but there also appears to be cruciform structure bounded north and west by a curved line. Its purpose is not obvious, and it is not apparent in magnetometry, but it may be obscured by plough marks. 23 Figure 4-10 Annotated twin probe resistance A few grids were selected to be repeated using the RM15 with 1 m probe separation. This was to see if any depth information, particularly the depths of the inner ditch and the northeast cordon, could be ascertained. The areas subject to this means of survey were shown in Figure 3-5 which also showed the location of radar trials. Two areas were measured in this way. Three grid squares in a line were surveyed thus along the northern inner ditch, and a square block of four grid squares was surveyed along the northeast cordon ditch, particularly aiming to include the central gap. The plot of the inner ditch is shown in Figure 4-11, and shows little that cannot be seen in the full resistance plot. It is unfortunate that we were not able to observe any more structure in the ditch, but at least the pattern indicates a deep ditch, greater than the metre nominal capability of the RM15 in this mode. Measurement of the northeast cordon ditch provided some extra information. The ditch appeared again to be very deep, but the gap appears to be irm ground all the way down, and the north and south ditch sections appear to have rounded ends. The low resistance heading north-east probably relates to a plough lynchet. Figure 4-12 shows comparison of the plots with 0.5 and 1 m probe separation. The dark spot in the southern ditch section terminus is a mediaeval bound stone. This is still visible on the surface, and is indeed distinctive in that it is a double stone, shown in Figure 4-13. 24 Figure 4-11 Inner ditch with 1 metre probe separation Figure 4-12 .5m probe separation left and 1m separation right Figure 4-13 Double boundary stone in northeast cordon ditch 25 4.4 Ground-penetrating Radar Three areas were surveyed by ground-penetrating radar (igure A-5). The irst was to the south, over the area of the large elliptical feature visible on magnetometry (see igure 4.3[6]). This feature was assumed to be shallow, so the 500 MHz head was used. The radar added no extra detail in this case. The area of the ellipse was noted to be a slight hollow during the survey. Its was only typically 100 mm below expected depth, too shallow to show on a contour survey at 1 m intervals. The other two areas were surveyed using the 250 MHz head. The irst area was the central 20 m section of the three grid squares surveyed with 1 m probe separation along the northern inner ditch. Only 18 m was surveyed north-south, as the inal section involved a steep slope up to the top of the rampart.. The start of the rampart slope was included, but it was considered that the slope might be enough to distort the plot. The inner surface, the ditch and the base of the rampart could be seen, even at nominal depths of 1.5 m (wavespeed was uncalibrated, but assumed to be 0.8 m/ns). Otherwise, little extra information was gained, other than conirming that the ditch was deep. Depth slices are shown in Figure 4-12. The third area corresponded with the 40 m square subject to resistance survey at 1 m probe separation over the northeast cordon ditch. Depth slices are shown in Figure 4-13. This survey proved useful in showing the ditch formation to extend to depths of over 2 m nominal, although the depth would be overstated if the wavespeed were less than the 0.8 m/ns assumed. 1.25 metre 1.5 metre 2 metre Figure 4-14 Inner ditch depth slices 26 1.5 metre 2.0 metre 2.5 metre 3.0 metre Figure 4-15 Northeast cordon ditch depth slices 4.5 Contour Mapping Contour mapping using an edm-type device was not possible, as intervisibility was so poor, and the survey need to extend down the hillside below the ramparts to show the steepness of the ascent. We are very grateful to Chris Ellis for volunteering to do the survey for us using high precision GPS equipment, and to Wessex Archaeology for the loan of the equipment. The hill top is devoid of tree cover, except on the northern face well below the ramparts, so good results could easily be obtained. Some 9000 readings were taken, including the ramparts, the areas below the ramparts, and along the lynchets (above and below) as well as many general hilltop readings. The data can and will be used in detailed discussions in the following section, but an example shown here shows the whole area plotted in 1 m contour intervals, as demonstration of the data quality. This is shown in Figure 4-16. 27 Figure 4-16 Precision GPS contour survey 1 m intervals Detailed area studies are shown with contours plotted at 0.2 m intervals. This gives good detail of earthwork, and indicates areas of rough ground below the ramparts, but these areas would need much higher data density to give a comprehensive picture of the quarrying and slumping. Figure 4-17 shows the southern end of the hillfort. The remaining portion of ditch can be seen by the dark band, where the contours reverse, above the steep slopes where the contour lines almost merge to a uniform black. The lynchets show up quite clearly, too. Figure 4-17 Southern end of hillfort .2 m intervals 28 Figure 4-18 shows the hilltop, to the north of Figure 4-17. The lynchets stand out quite clearly, as does the northern rampart, where it survives. Use of high resolution contour survey serves here as an alternative to earthwork survey, with the added beneit of exact national grid locations. A slight depression is also evident at the location of the large elliptical anomaly discovered in the magnetometry survey (see Figure 4-4[6]). The steep sides appear as dense black contouring. Figure 4-18 Central hilltop .2m intervals 29 Figure 4-19 shows the northeast corner. The contours pick out secondary mounds outside the ramparts around the extreme corner. Diagonal cuts represent footpaths rising to the top of the ramparts, but these may be modern. Regular straight contour lines probably indicate where there is insuficient data, where it was too steep to take the device safely. Figure 4-19 Northeast corner .2 m intervals 30 Figure 4-20 shows the northwest part of the hillfort, and the main entrance can be seen as a depression between the ramparts. The steep slopes resulting from quarrying are also evident. Figure 4-20 Northwest corner and entrance .2 m intervals 31 The southeast entrance is at the centre of Figure 4-21, and is visible as a discontinuity in the steep slopes. There is also a splaying out of the contours just at the top of the ramparts, but this may have been caused in part by modern footfall. Some slumping of the rampart is visible going off the top of the plan, and rough ground below the ramparts is discernible generally. Figure 4-21 Southeast entrance .2m intervals 4.6 Synthesis Both magnetometer and resistance plots were very lively, and many features could be seen in both. There were a few features which could only be seen by one of the technologies, but the two methods were generally complementary. The features most obvious in both were the plough lines and lynchets from the agriculture of mediaeval times to the nineteenth century. This section attempts to ignore those marks to describe the earlier archaeology. In some cases, this is not possible as some features could be attached to any period, and their context cannot be deined clearly. The subjective nature of this interpretation has to be made clear. 32 Figure 4-22 is an attempt to pick out the early occupation features of this hilltop. The circles are taken to represent dwellings. However, the large ellipse at the south, marked in mauve, would seem too big to be a house. It may be the remains of a barrow, in which case it would be facing barrows on Bathampton Down on the other side of the valley (Thomas, 2008), or it may some form of working surface. It showed clearly in magnetometry but not in other instruments. Figure 4-22 Possible occupation features The dwellings, blue circles, seem to be in clusters on the sides of the domed hilltop, principally on the eastern side, which would be slightly more sheltered, but with less of a strategic view. Where there are multiple overlapping circles, it is not generally possible to provide any chronology. There seem, nonetheless, to be some 30 sites of houses which could have been in use at the same time, although geophysics cannot provide any irm evidence for this. Indeed, the marked sites are not necessarily the complete set of houses; there could easily be others which could not be distinguished from the later archaeology. There are some approximate alignments of houses, but there is no obvious ordered street pattern. There appear to be funnelling lines emanating from the north-west entrance, and there are possible gate-houses on the ramparts. Similar lines appear to emanate from the south-east entrance, although it is just possible that these may coincidentally relate to mediaeval features. However, they do seem to be directed at the most prominent of the huts, which stands on the summit. The approach to this entrance from outside is very steep, and this would have limited the use this access. However, it does suggest that this entrance was original, not a later cut, as had been assumed. 33 The north-east corner is certainly curious. There are a couple of possible circles within the cordon ditch but generally there appear to be patterns of multiple post holes, although these do not form into regular rectangles. It is still most likely to suggest raised buildings such as granaries. The possible dew pond in the north-east extremity also suggests this area was for provisioning rather than dwelling. The cordon ditch separates it from the main hilltop, but it is not a secure cordon. There are some features on the main hilltop which could be pits for storage, but it is dificult to be sure that these are not anomalies relating to later ploughing.. Resistance shows a feature which might be a well near the north ramparts, and shows a set of ine lines cut into rock near the centre. These form a pattern which approximates to rectangular. These could possibly represent beam slots, but they do not form a regular enough pattern for a building. There is also a small circular feature. They are close to the most prominent of the huts, and may relate to it, but geophysics cannot elucidate their use. The inner ditch appears as a very strong feature around the whole periphery of the hilltop except in a few places where the edges have tumbled away. This suggests it is rich in organic remains, but it would also be true that soil from dwelling and also from later agriculture would have migrated off the hilltop and lodged in this ditch. It does tend to suggest high levels of activity nonetheless, and the magnetic susceptibility plot also indicated high levels of magnetically enhanced soil. This, along with the hut circle signals suggests that Solsbury Hill was a well-occupied settlement, not just an occasional sanctuary. Figure 4-23 Decorated Solsbury pot sherd. Decoration marked out by W.A. Dowden excavators. Courtesy UBSS. 34 5 The Hillfort in its landscape Solsbury Hill has similar geology to Charmy Down, which overshadows it from the north, and is probably an outlier which has become detached from it, so that there is now a deep narrow valley which separates the two hills. Deeper stream valleys separate it from Bannerdown on the east and from Lansdown on the west. All of these are a little higher than Solsbury Hill and delect the worst of the weather away. To the south, across the river valley is Bathampton Down, slightly higher than Solsbury, and with Iron Age ramparts on top (Thomas, 2008). These hills all have very steep sides, rolling off at the top to apparently lat plateaux, although Solsbury and Bathampton are deinitely dome-shaped when observed in detail. Because it is an individual hill, with steep slopes on all sides, it is an obvious candidate for a fortiied or protected settlement site, especially with weather a little more benign than might be expected. It also has springs just below its summit, so potable water is close at hand, and there are even features detected by geophysics which could be wells within the ramparts. Just beyond the point on the river where it passes between Solsbury Hill and Bathampton Down, the valley splits, with the Bybrook continuing to the east, whereas the Avon heads south to reach Chippenham by a more circuitous route. This leaves Solsbury Hill with views east towards Oldbury, Cherhill, beyond Chippenham, and also south in the direction of Westbury and Salisbury Plain. As the eastern side of the Avon valley beyond Brown’s Folly drops away in height, the northern edge of the plain and Bratton Hillfort are visible above these hills. To the southwest, Solsbury Hill overlooks Bath, and the Mendips over the top of the hills behind Bath (Southdown). From west through north to east, the view is very much restricted by the higher hills of Landsdown, Charmy Down and Bannerdown, although there is a gap just below the plainly visible earthworks of Freezing Hill. Figure 5-1 shows the physical geography of its location, and Figure 5-2 shows the viewshed from Solsbury Hill. The viewshed used here is composed from three sectors at rampart height. The view from the summit is slightly more far-reaching, but is very restricted locally. This will be discussed later in the chapter. Figure 5-2 also annotates a number of other hillforts and Iron Age settlements within its catchment. Only a limited number are visible. Figure 5-1 Solsbury Hill in the local landscape 35 By its very nature physically, the hilltop was ideal as a fortiied retreat and stronghold, and by its placing it is fortuitously at a very good location to dominate movement along two strategic river valleys, so it is an ideal candidate for fortiication. Figure 5-2 Solsbury viewshed The viewshed provides one way to quantify its effectiveness, but this is also expressed by showing the views from the hill, here producing panoramas. These include the panoramas from the ramparts and also from the centre of the hilltop, showing how they differ, and also a ‘reverse panorama’, looking back at Solsbury Hill from various locations to show how it is perceived within the landscape. The views from the hill are spectacular, and it would be possible to show panoramas of interest from any point around the edge of the hill. The view from the middle of the hilltop is restricted to only the higher close hilltops and the distant ranges to the east, south and west. For our purposes, four panoramas are included so as to show the maximum exposure of the closer valley loors, which would have been of more immediate interest to the inhabitants. The view from the Batheaston entrance, which we believe dates to the Iron Age, overlooks the St. Catherine Valley, the Bybrook Valley, and the Avon Valley as well as Bannerdown, where the Fosse Way heads north, the hills near Avebury and Devises, and Bathford Down. The River Avon is visible in places from this side in spite of currently being tree-lined. The northeastern corner has the most obscured view because of the tree cover at the bottom of the steep slope at the northern end of the hilltop. It does provide a view further up the St. Catherine’s Valley and also of Chilcombe Bottom and Charmy Down, the plateau just to the north of Solsbury. There is evidence of early occupation at Charmy Down and Iron Age ield systems (Grimes, 1960, 200-231). 36 The signiicance of the viewshed should be perhaps be seen in light of Little Solsbury’s position in the Dobunni tribal territory. The Dobunni occupied both sides of the Severn Valley encompassing most of modern Gloucestershire and North Somerset (Corney, The Roman Villa at Box, 31). While it is impossible to know the actual history of tribal interactions in that period, Solsbury offers visual contact with the northern edge of the Atrebates territory. As a prominent stronghold near the tribal boundaries, it could have been an important outpost for trade, intelligence and communication. Figure 5-3 Solsbury spearpoints. Courtesy UBSS Panoramas The viewshed only partially illustrates the visual situation of Solsbury Hill. In reality the dome shape of the hilltop restricts views from the centre and from various points around the edges. From the centre, all of the closer areas of the valley bottoms disappear. From the viewpoints at the entrances and other corners of the rampart, the dome blocks views of the opposite sides. The panoramic photographs following show the practicalities of these restrictions. Even though Little Solsbury occuppies a magniicent defensive position, keeping watch over the surrounding countryside would have required some organisation and manpower. 37 view to NE from centre view to NW from centre view to SW from centre view to SE from centre view from NE corner view from NW entrance view from SW corner view from SE entrance 6 Comment and Conclusions Solsbury Hill has proved to be a much busier place than previously thought. Speculation on the occupation of the hilltop was based only on evidence discovered by Antiquarians, a small survey, and a series of small excavations in the 1950s. A good deal more has been learned about hillforts since then, and geophysics provided the opportunity to learn much more about Solsbury Hill in a nondestructive manner. Dowden concluded that the occupation was primarily around the periphery of the plateau just inside the rampart based on his small sample (Dowden 1957, 23). This proves to be inaccurate, and most of the occupation would appear to be well into the centre of the hilltop. The number of hut circles indicates a large population compared to many hillforts (Wessex Hillfort Project, Payne, 143) and would suggest Solsbury was a well established community. He also came to the conclusion that there was no inner ditch inside the rampart (Dowden 1957, 26). Our results contradict this theory. There is clear evidence for an inner ditch around the entire hilltop except for where the edge has been completely eaten away by quarrying. Figure 6-1 Rampart trench Dowden 1958 According to Dowden, the rampart was 20 feet wide between the two retaining walls (Figure 6-1). The outer wall was estimated to be at least 12 feet high (Dowden, 1957, 27). If this were the case, Solsbury in the Iron Age would have been a most impressive site. The amount of quarrying and soil removal that has taken place is remarkable (see Appendix D). The most interesting new large feature we found was the northeast cordon ditch with an entrance into a separate precinct. This section does not appear to have any round houses, so it could be possible that it is an area set aside for grain and food storage. There is evidence of numerous pits and/ or postholes which would support this idea. One of the main functions of a hillfort was as a distribution centre for food (Wessex Hillforts Project, Cunliffe, 154), and Solsbury could have served this purpose for the area north of the River Avon in this vicinity. For example, Charmy Down just to the north shows extensive Iron Age ield systems but doesn’t have any signiicant settlement (Grimes, 1960, 200-231) and could have easily been part of a Solsbury distribution system. River terraces along the Avon would have also had extensive ields. Excavation in the northeast corner of Little Solsbury could supply some detail to the distribution centre theory. The cordon ditch could be up to three metres deep, so inding undisturbed evidence is quite likely. 40 Another area to clarify is the actual chronology of the hillfort. The basic dates arrived at by Dowden it the general time lines of hillforts in southern England, though on the shorter time scale of his era. There was a period around 300 BC when most hillforts show signs of upheaval (Wessex Hillforts Project, Cunliffe, 160). Many were abandoned at this time, and the surviving hillforts were strengthened into what are called developed hillforts with stronger defenses. The Solsbury rampart seems to have come down during this period, but the defenses weren’t replaced , although there seems to have been substantial occupation for up to 200 years afterwards. The inal abandonment of Solsbury took place earlier than general period of hillfort abandonment in the irst century BC and early Roman period, which raises further questions. Dowden’s chronology was based on the dating evidence available in the 1950s, but Iron Age dates have been extended much further back in time since then (Wessex Hillforts Project, Cunliffe, 155). The exact date of the rampart coming down would help us to understand where Solsbury its in with the developed hillforts of the later Iron Age and possibly shed light on the Batheaston entrance question. It is generally assumed to be of a later date than Iron Age. Early Iron Age hillforts typically had two opposite entrances, and one was usually blocked when the hillfort defenses were strengthened in the later Iron Age (Wessex Hillforts Project, Corney, 138). Solsbury seemed to be an early-middle Iron Age hillfort with one entrance. Our data would seem to indicate that the Batheaston entrance has been in use for a very long time, and perhaps the removal of entrance features at an early date obscured the visual evidence for a typical Iron Age entrance. This question can only be inally answered by excavation. Figure 6-2 Batheaston Entrance Very little, if any, evidence of earlier features was found by our survey. The centuries of ploughing have no doubt obscured or destroyed any more ancient features not removed by the Iron Age inhabitants. Much of the hilltop has less than 35cm of topsoil, so continuous cultivation would soon eliminate any traces of more subtle usage. The exception could be the large elliptical feature on the southern end of the hill (Figure 4-22). This might either be an earlier barrow or a large, ostentatious round house of the late Bronze Age-early Iron Age. Again, excavation would probably supply dating evidence. The current trend in hillfort studies is to try to understand the wider landscape and activities occurring outside the hillfort (Wessex Hillforts Project, Payne, 139). Solsbury’s standing in the immediate 41 area, relationships with other nearby population centres, such as Bathampton and Bury Camp, and status in the tribal territories of the Iron Age are all subjects about which we know very little. BACAS plans further geophysical surveys down the southern slopes (Figure 6-3) of Solsbury to look for possible earlier remains and more Iron Age usage. From the wealth of material found previously in the quarries and during the excavations (Falconer, Dowden), it can be assumed that there are many more inds in all the ditches and deeper soil on the hilltop. Since we have located these ditches and numerous other features, it would be very fruitful to make further excavations at some date in the future. Excavation could possibly throw light on these questions. There is much more to uncover on Solsbury Hill. Figure 6-3 Little Solsbury from hot air balloon over Lansdown courtesy of the Batheaston Society 42 Bibliography References: Andrew Payne, Mark Corney and Barry Cunliffe. The Wessex Hillforts Project. ©English Heritage 2006. Collinson, John and Rack, Edmund, 1791. “The history and antiquitities of the county of Somerset.” Vol. 1, 99-110. Corney, Mark. The Roman Villa at Box. ©KOBRA Trust 2012. Dowden, W.A. “Little Solsbury Hill Camp. Report on the excavations of 1955 and 1956” UBSS Proc. Vol. 8 (1) 1957 18-29. Dowden, W.A. “Little Solsbury Hill Camp. Report on the excavations of 1958” UBSS Proc. Vol. 9 (3) 1962 177-182. Evans, John. “On a discovery of Flint Arrowheads and Other Stone Implements at Little Salisbury Hill, near Bath”. Transactions of the Ethnological Society of London, Vol. 4 (1866), 240-243. Published by: Royal Anthropological Institute of Great Britain and Ireland. Stable URL: http://www.jstor.org/ stable/3014291. Falconer, J.P.E., and Adams, S.B., 1935. “Recent inds at Solsbury Hill Camp near Bath” University of Bristol Speleological Society Proceedings, Vol. 4 (3), 183-222. Green, G.W. 1992. British Regional Geology: Bristol and Gloucester Region. Third Edition. British Geological Survey/NERC. HMSO – London. Green, G.W. and Donovan, D.T. 1969. The Great Oolite of the Bath area. Bulletin of the Geological Survey of Great Britain, No. 30, 1-63. Grimes, W. F. Excavations on defense sites. London, HM Stationery Ofice, 1960, 199-235. http://mapapps.bgs.ac.uk/geologyofbritain/home.html Thomas, R. 2008. A Sacred Landscape. The Prehistory of Bathampton Down, Bath. W. G. Collins and T. C. Cantrill, 1909. “Solisbury Hill Camp, near Bath.” The Antiquary, Vol. 5, 326-331, 419-425, 451-456. W. Phelps, 1836. “The History and Antiquities of Somersetshire”, 102-103. 43 Appendix A. Details of gridding and surveying A1 Magnetometry The magnetometer was a Bartington 601-2 dual luxgate gradiometer, taking readings at 4 per meter along traverses 1 m apart, giving 1600 readings per 20 m grid square. Grids were started in the south-west corner, the irst traverse heading (grid) north. The irst traverse was 1 m west of the west line between corner posts, the last was between the east corner posts. Traverses started 0.25 m north of the line between south corner posts and inished on the line between north grid posts, in accordance with Oswin, 2009, p115 igure 5.8c. The survey started at the northwest corner, and continued eastwards along the line of grids, then inilled any grids to the north of this line, and then continued heading south in an orderly sequence of west-to-east lines. The grid map is shown in igure A1. All data were downloaded using Bartington proprietary software, so that they were already sorted into a series of 20 lines of parallel data. Downloaded grids were passed through BACAS proprietary zero-median de-stripe software before importing into INSITE. The raw data iles are preixed ‘M’, the de-striped data (also with any dummy readings converted to ‘NUL’) are preixed ‘D’. These ‘D’ iles were used to generate the plan. Figure A1 Magnetometry grids A2 Magnetic susceptibility The readings from the MS2 were recorded on paper along with the eastings and northings given on a handheld GPS kit. This gave coordinates to a precision of 1 m but was only accurate to 5 m, and the latter is taken as the maximum resolution of the plot. All readings were transcribed to XL spreadsheet as column 1:easting; column 2: northing; column 3: susceptibility. The data ile was then 44 copied into Dplot freeware as XYZ data to produce a contour map. Shaded contour bands were used for each rise of 10, and this was capped at 16 bands. A3 Twin probe resistance The resistance survey used the same grid squares as magnetometry and were done in a very different order. They generally applied the same rule of start in the south-west corner heading north, but with only 2 readings along each traverse, traverses 1 m apart, giving 800 readings per 20 m grid square. Although this was the general rule, there were a few grids started in error at the north-west corner, heading south. These are identiiable by the arrow position and direction in the squares in igure A2. Both TR/CIA and Geoscan RM15 devices were used, and downloaded via BACAS proprietary software before being taken into INSITE. All resistance grids are preixed ‘R’. Use of the two machines expedited the survey but caused problems in that the TR data are downloaded ready to sorted to parallel, whereas the RM15 data are downloaded in zig-zag fashion. Both types had to be imported separately into INSITE, but they could be mapped onto the same survey plan. In igure A2, the TR data are indicated by a blue arrow, and the RM15 data are indicated by a red arrow with a cross bar. Figure A2 Twin Probe Resistance Grids The numbering sequence is not orderly as was the case with magnetometry. In April, there was an initial attempt to complete at least the periphery of the hillfort, but even this was not completed. Once the survey was re-started, grid squares were illed in working from those points where the grid could be reconstructed by triangulation from ixed points, and as grid posts left unattended for periods were liable to be uprooted either by cattle of by visitors, so only small areas were gridded out at any time. There was also an added requirement to keep the two machines well separated when in use together to avoid crosstalk. 45 After any long periods without surveying, the numbering sequence would be re-started, so there are gaps of a few numbers from time to time. For instance, if one run had inished on grid 133, the next might start at 141 rather than 134. Some grids were repeated if the quality of the original data were not good enough, so there could be some sequences interrupted by very different grid numbers. There were a few grids, particularly numbers 157, 162, 163, 164, and 165 where the apparent resistance was measured much higher than normal, as a cable connector fault had gone unobserved. Many of those grids were repeated, but there are a few that still gave good quality data, and these were retained and are listed above. In a few grids round the periphery, controls were pressed in the wrong sequence to end lines early and start new partial traverses. These were repaired manually within INSITE. These caveats suggest poor data control, but there are relatively few cases enumerated here, given the extent and time period of the survey, and overall data quality was very good. Two small areas were measured using the RM15 with 1 m probe separation, Their grid numbers followed on directly from the end of the main survey. Figure A3 shows the grids for the measurement along the inner ditch, and igure A4 shows the grids measured over the northeast cordon ditch. The locations of these extra surveys were shown in igure A-5. Figure A3 Inner ditch grids Figure A4 Northeast cordon ditch grids A4 Ground-penetrating radar The three areas subject to radar survey are also shown in igure 4.10. Area 1 used a 500 MHz antenna, taking readings every 0.05 m, governed by an odometer wheel. Traverses were 1 m apart and they were surveyed in zig-zag format, so it is necessary to reverse the direction of every second traverse. One traverse (17) gathered no data and a replacement traverse had to be taken, so this has to be omitted while importing the data iles, and the numbering sequence changes then from odd up, even down, to even up, odd down. The sequence forms a square of 40 lines each 40 m long. Area 2 comprises 20 lines each 18 m long, using the 250 MHz antenna with readings every 0.1 m. Again every second line needs to be reversed. Area 3 also used the 250 MHz antenna reading every 0.1 m. There are 40 lines of data each 40 m long, with each second line needing to be reversed. In all cases, traverses were (grid) north-south, with traverse 1 being towards the north. 46 A5 GPS survey The data were automatically recorded by the Leica GPS device and downloaded to a spreadsheet. Grid references were six-igure, followed by three decimal places, and height data were also given to three decimal places. The data were in three columns, easting; northing; height, with additional columns providing commentary. For use in producing the contour maps, the irst igure was stripped off the six-igure northing and six igure easting to provide data compatible with the magnetic susceptibility data. The stripped off preixes were equivalent to stating that the site was in National Grid square ST, so that, for example, a reading of 376700, 168600 could be interpreted as ST7670068600. Appendix B. LIDAR B1 Geomatics LIDAR Because of the tree cover around the north end and thick turf on the summit obscuring possible features, we requested and were kindly supplied with a LIDAR image of the hilltop. According to the Geomatics website, LIDAR (Light Detection and Ranging) is an airborne mapping technique, which uses a laser to measure the distance between the survey aircraft and the ground surface, including buildings and other assets (above ground pipelines, highways, street furniture, power lines, railway tracks), as well as vegetation. The data obtained is processed into georeferenced coordinates and can be displayed with structures and vegetation removed. For further information, go to the Geomatics website: https://www.geomatics-group.co.uk/geocms/homepage.aspx 47 Appendix C. Views of Little Solsbury Little Solsbury Hill is not an isolated hill in a low landscape. It is surrounded by other hills that have all been created by the erosion of the limestone plateau. Little Solsbury is actually slightly lower than the surrounding hilltops. The valleys between the hills have steep enough slopes to make access to the tops of Solsbury and the other hills a long climb. Little Solsbury is however unique in this area in that it is cut off from the other hills on all sides, thus providing a better natural defensive position. As is shown in these views, it is a long climb to the top from any direction. Also, the closer you are to the hill, the more impressive the view. When Little Solsbury had a 12 foot high stone rampart around the entire top, it would have been the most impressive structure in this part of the Dobunni territory. Little Solsbury from Lansdown on the northwest. Lansdown is the hill starting in Bath and going to the north. From upper Larkhall on the west. Larkhall is a former village, now part of Bath. 48 From Larkhall on the southwest. The hill is prominent in Larkhall. From Alexandra Park in Bath on the southwest. Solsbury is three miles from the centre of Bath. 49 From Bathwick Hill on the south. Solsbury blends into the other hills from this direction. From Bathampton Down on the south. The top of Solsbury is prominent from the area of the Iron Age settlement on Bathampton Down. 50 From the Avon Valley by Bathford on the southeast. The hill stands out when seen from the upstream direction of the River Avon Valley. From London Road Batheaston on the east. Solsbury looms over much of Batheaston. 51 From Northend Batheaston on the northeast. Northend is built on the lower eastern slope of the hill. From Charmy Down on the north. This is the closest neighbouring hill, and it had extensive Iron Age ields. Although Charmy Down is the closest hilltop, there is a steep sided valley, Chilcombe Bottom, between the two. 52 Appendix D. Batheaston Freeholders Guidelines The casual approach to quarrying in rural areas is best illustrated by this1950s brochure (Figure D-1) from the Batheaston Freeholders. Quarrying continued up until at least 1968 when Denis Grant King found a collapsed section of rampart and ditch caused by quarrying of the oolite layer lying below on the north end of the hilltop. He found a good deal of pottery, bone, and slag in the debris. I hasten to add that these guidelines predate the National Trust ownership of the hilltop and that in our era the Freeholders are an active conservation group and were very helpful with organising our work on Solsbury Hill. Figure D-1 Batheaston Freeholders rules for quarrying Appendix E. Details of gridding There are eight permanent reference points measured in to reconstruct the grid pattern. These consist of two each grid square corners on the north, south, east and west sides of the hilltop according to where we found appropriate permanent features. We recorded the distances and photographed where the measurements were taken on the particular permanent features. Grid Points South Point 1 Memorial Stone NW corner West to point 35.19m Trig Point N corner NE to point 62.04m 53 Figures E1-2 Measuring points for 1st southern grid reference South Point 2 Trig Point N corner NE to point 36.67m Memorial Stone NW corner West to point 75.24m Figures E3-4 Measuring points for 2nd southern grid reference West Point 1 Field stone South corner South to point 80.20m Flat Stone SE tip South to point 81.54m 54 Figures E5-6 Measuring points for 1st west grid reference West Point 2 Flat Stone SE tip South to point 44.17m Field Stone South corner South to point 40.94m Figures E7-8 Measuring points for 2nd west grid reference NE Point 1 Field marker East corner East to point 24.72m Double Field mark NE corner East to point 39.94m Of SW stone 55 Figures E9-10 Measuring points for 1st northeast grid reference NE Point 2 Double marker South corner South to point 11.80m Field Marker West corner SW to point 16.27m Figures E11-12 Measuring points for 2nd northeast grid reference N Point 1 Large lat Stone NE corner NE to point 50.30m Rampart marker stone grooved West to point 11.30m 56 Figures E13-14 Measuring points for 1st north grid reference N Point 2 Rampart marker stone grooved West to point 31.15m pull tight Large lat stone NE corner NE to point 39.20m Figures E15-16 Measuring points for 2nd north grid reference 57 Little Solsbury Hill from Bathampton