The complex of Roman buildings excavated by MAAG at East Farleigh, 2005-17: an interim report

129 the complex of Roman buildings excavated by MAAG at East Farleigh, 2005-17: An interim report stephen clifton In 2005 Maidstone Area Archaeological Group (MAAG) was asked to reexcavate a Roman building discovered in 1839 on land adjacent to the River Medway in East Farleigh. Twelve years later the group had uncovered at least six buildings, under the direction of the late Albert Daniels with the assistance of the author and other group members. Some parts of the site proved to be remarkably well preserved and represented a number of different phases of activity spanning the late Iron Age to the 5th Century ad. Expectations were that the Group were investigating ‘just another Roman villa’, but these were confounded, and the site is now confidently identified as a religious sanctuary, with buildings of unusual architectural form. In the 1830s a Roman building was discovered by workmen on agricultural land in East Farleigh whilst putting in new trackways and agricultural buildings. A measured plan was published by J. Smith in 1839 titled ‘Foundations of a Roman Villa at East Farleigh’, and is the first reference to Roman buildings on the site (Fig. 1). It also refers to previous removals of ‘foundations’ (Smith 1839, 57). MAAG were invited to the site, centred on NGR TQ 72850/53620, by the modern owners in 2005 to see whether there was any more to the building and to improve on the rather basic plan. When the group started to excavate, under the direction of the late Albert Daniels, latterly assisted by the author, there was very little expectation of finding anything other than a previously disturbed site. A resistivity survey was carried out, followed by targeted test pits, and the outline of a building was established. Despite the lack of preservation in some places it soon became clear that the site had not been examined archaeologically, and that it was largely untouched. The building (Building 1) was explored with selective trenching over two seasons that identified the layout of the structure and some tentative dating. However, it did not match the building on the 1839 plan. Indeed, during the group’s time at the site this structure was never identified. It seems most likely that it was largely removed to make way for the hop pickers’ accommodation which is in the south-east corner of the site; the brick wall footings of the modern buildings and a concrete yard were still evident. MAAG went on to investigate the site for a further twelve years, uncovering at least six buildings in total as well as identifying significant Iron Age activity. The group are continuing to excavate on land to the south of the buildings where evidence for further activity before and after the Roman period is emerging (Clifton 2019). STEPHEN CLIFTON 130 To appreciate the site at East Farleigh it is important to have a broader understanding of the local Roman geographic context (Fig. 2). Along this part of the river Medway there are a number of known villa sites, at Teston, Barming, and other buildings in Maidstone, such as the Mount Roman villa and another potential villa to the east, near Loose Road. There are Roman-period burials along the route of the modern roads in East Farleigh, which indicate the existence of a similar road pattern in the Roman period and suggests that there may have been a river crossing at East Farleigh where the buildings excavated by MAAG look across the valley at the possible villa settlement at Barming (Payne 1880, 169). It is easy to imagine a crossing point here, probably predating the Roman era, hinted at by the late Iron Age finds and features found on both sides of the river. Unfortunately, none of these nearby villas have been fully excavated, with the exception of the Fig. 1 Plan of the 1839 Roman building discovered at East Farleigh (Smith 1839). ROMAN BUILDINGS EXCAVATED BY MAAG AT EAST FARLEIGH: AN INTERIM REPORT 131 Mount (Houliston 1999), which sits to the east of a bend in the river Medway and is probably associated with the Roman road from Rochester to the south coast (Margary 1946, 33). Iron Age Origins The earliest activity close to the East Farleigh buildings appears to date from the late Iron Age, based on the pottery (Lyne 2019) and two coins dated to 10 bc and ad 1 found in later contexts (Holman 2019). There are two ditches (A and B), that underlie several of the Roman buildings (Fig. 3). These were traced over a 70m length, running roughly east-west in parallel, approximately 5.5m apart and then turning to the north where it was possible to glimpse at least one of the ditches beneath Building 6 and other features. Pottery from the ditches has suggested that they remained open for some time, possibly as long as 150 years in the case of ditch B. Several coins have been found in contexts possibly related to the cessation of the ditches dated to the late first or early second century ad and are consistent with four cremation burials found approximately 150m to the south in 2019, (Clifton 2019). Unfortunately, it was not possible to explore the full extent of this ditch system, and so we do not really know its purpose or significance. The fill of the ditches at East Farleigh indicated a continuation of use well into the Roman period, yet we did not encounter any further evidence of settlement elsewhere on the site that could be dated to this period in support. One significant find in ditch B was an almost intact jar in black Belgic grogtempered fabric, dated 50 bc-ad 60 (Lyne 2019). It was found in one of the lower Fig. 2 Map of Roman north Kent (redrawn, after Andrews 2001). STEPHEN CLIFTON 132 Fig. 3 The Iron Age ditches at East Farleigh (Daniels and Clifton 2020). ROMAN BUILDINGS EXCAVATED BY MAAG AT EAST FARLEIGH: AN INTERIM REPORT 133 fills. Assumed to be a ritual deposition due to its complete nature, perhaps placed when the ditch was partially filled with water. Given the dating of the ditches and their position in the valley floor close to the River Medway, it is possible that we are looking at an oppidum, similar to Quarry Wood at Loose, which Detsicas suggests was part of a network controlling the river crossings, (Detsicas 1983, 2). Quarry Wood also has numerous later Roman structures associated with it. However, there are topographical limitations at East Farleigh. The site is situated on a plateau adjacent to the River Medway with a steep slope down to the river, limiting its potential extent. However, as has been shown at other similar sites, such as Verulamium, oppida can vary greatly in size, and there is much debate about the way that they functioned, with clear evidence for zoning and a wide variety of different activities (Niblett 2001, 47). The continuity of activity from the first century bc through to the Roman period, with no apparent hiatus around ad 43, is a common finding at Roman period sites in the South-East. Subsequent work by MAAG, 150m to the south, on a break of slope has revealed further activity from the first century ad (Clifton 2019). Here a possible Iron Age enclosure ditch is cut by a palisade trench with associated cremation burials dated approximately ad 70-100 (Lyne 2019). Once again, no identifiable domestic structures, or other indicators, were observed leading to a tentative working hypothesis of ritual usage. The Six Roman-Period Buildings discovered by MAAG The buildings were numbered by MAAG in the order in which they were discovered and excavated. Buildings 2, 3 and 4 turned out to predate 1, 5 and 6, hence the order in which they are described below (Figs 4 and 5). Building 2: the first of the stone buildings constructed on the site seems to coincide with the filling in of the Iron Age ditches c.ad 150. This is a barn-like structure orientated roughly east-west, with a large 3.27m wide door opening in the southern wall. It is estimated to have been just under 27m in length, but the width is less certain, as all of the northern side of the building appears to have been lost during the Roman period remodelling and then further truncated by the 19th-century trackway and revetment. Only the southern corners and the southern wall remain, removed to the level of the floor and truncated at the western end by the later building. Its shape would suggest a utilitarian building and it would seem likely that there were other buildings associated with it, but none were identified during the excavation programme. It is hard to date the demolition of this early phase of building; however, it would appear from the pottery evidence that the later building that replaced it (Building 3) was constructed around ad 200, suggesting a demolition date prior to that but after 150. Building 3: appears to have replaced Building 2, although also losing its northeastern corner to the modern revetment and remained partially standing to a level of eight or nine courses of stone in its south-western corner (Fig. 6). It measured 27 x 7.8m and was on a slightly different alignment to the building that it replaced. It had a large, 3.39m wide entrance in the southern wall. There were originally STEPHEN CLIFTON 134 two internal walls, and evidence for an external door in the northern wall of the western end room, but the corresponding position in the eastern end has been lost. There is evidence for a metalled surface overlying the demolished internal wall at the eastern end suggesting that the removal of the internal walls represented a repurposing of the building, rather than a preliminary stage of the demolition. At the western end, the floor consisted of the natural sandy Gault Clay which had been turned dark red from the heat of numerous oven structures of varying size dug into the floor. Pottery from the building suggests that this transformation in usage took place during the third century. The reddening of the clay surface stretched to halfway down the building but did not impact on the walls and was also confined to the inside of the building. It also crossed the area where both internal walls had been removed suggesting that it had occurred before the insertion of the ovens. Fig. 4 Site plan of the complex of Roman buildings at East Far leigh (Daniels and Clifton 2017). ROMAN BUILDINGS EXCAVATED BY MAAG AT EAST FARLEIGH: AN INTERIM REPORT 135 Fig. 5 Plans of Buildings 2, 3 and 4 (Daniels and Clifton 2017 ). STEPHEN CLIFTON 136 A later phase of usage was represented by three larger ‘corn-dryer’ type ovens which were inserted into the building, cutting through the reddened floor surface, and although these yielded no dating evidence, they must have been introduced late in the building’s life, probably in the fourth or fifth centuries. When excavated, a large capping stone was found placed on top of one of the structures. This stone showed no sign of the effects of heat or indeed any usage, and so the conclusion must be drawn that this was a ritual act at the cessation of its useful life, a phenomenon seen elsewhere on the site. Building 4: Building 3 does not appear to have been constructed as a stand-alone structure – indeed there are two walls running off to the north and the west from the north-western corner of the building (Fig. 5). The wall heading north connecting to what appears to be the corner of another building, with an additional wall running off to the west from this corner. Unfortunately, this structure, Building 4, appears to have been all but lost to us, due to the 19th-century track and revetment which has left only a few stones at its south-west corner. Indeed, the revetment is built of stone suspiciously like that of the Roman era structure that it replaced. The wall running off to the west from the remains of Building 4 appears to have been demolished to accommodate the construction of Building 5 (see below), Fig. 6 The central area of Building 3 (Clifton 2011), looking north showing the reddened earth from the heat of ovens at the western end and the late corn-dryer inserted over the fill of ditch B. The southern wall and entrance to Building 2 can be seen in the centre, reduced to floor level. The southern entrance to Building 2 can be seen in bottom right . ROMAN BUILDINGS EXCAVATED BY MAAG AT EAST FARLEIGH: AN INTERIM REPORT 137 as it peters out a few feet from that building’s eastern wall. This would suggest that Building 3 was part of an interim phase of construction and there may have been other buildings associated with it that were either demolished during the remodelling of the site in the second or third centuries, or during the subsequent 19th-century ground works, such as the fragment of Building 4. No trace of any other buildings from that phase was found during the excavation. Whatever buildings were originally associated with it is clear that Building 3 was retained as part of the next phase of building, sometime in the first half o f the third century. Building 5: is one of four buildings that shares a geometric alignment around an apparently open courtyard space facing north-east. It is the most complete building and survives to waist height in places and retains its whole floor plan. It sits in the south-western corner of the site and is aligned with Buildings 1 and 6 (see below) which look to have been conceived as an architectural whole. The building measures 13.9 x 12.05m and has a corridor or ambulatory on three sides, and a central double cella (Figs 7-9). The corridor is 2.2m wide. There Fig. 7 Plan of Building 5 (Daniels and Clifton 2010). STEPHEN CLIFTON 138 are indications that the corridor had two partition walls, which may have been removable. The section of corridor on the southern side was separated from the rest of the corridor by a wall with a doorway. The floor of this separate area was different to the rest of the building having no obvious floor surface other than the local clay and hassock and appeared to be of a utilitarian function. The outer walls of the building had been removed down to one or two courses of stone, especially at the north-eastern corner, although the western wall retained stonework to approximately a metre in height. The inner cella was still standing to approximately 1.5m in places, although the north-eastern corner was severely reduced. The building was constructed of local ragstone with some tufa quoining evident. Unusually for Fig. 8 Drawing of Building 5 showing arrangement of internal d oors (Clifton 2020). Fig. 9 Buildings 3 and 5 seen from the south (Clifton 2020). ROMAN BUILDINGS EXCAVATED BY MAAG AT EAST FARLEIGH: AN INTERIM REPORT 139 Fig. 10 (above) Building 5 seen from the west and (right) painted plaster on external wall (Clifton 2010). Britain there was some painted wall plaster remaining on the external western wall comprising black banding separating a lower red section and an upper white/ pale blue section (Fig. 10). This plasterwork did not appear to extend around the corners to the adjacent walls. There was also evidence for painted plaster on the internal walls although none remained attached. This plasterwork was painted a deep ‘Pompean’ red, except for areas at the western end of the northern corridor and the southern section where it was painted white. There were no indications of any STEPHEN CLIFTON 140 windows, nor were there any remains of columns or column bases found. The western end of the building and the central cella were covered in a thick layer of tumbled stone, indicating that this area of the building had been standing to a significant height, until it either fell or was pushed over. There was no wall plaster amongst this stone layer indicating that it had probably fallen off due to weathering long before the wall itself collapsed. Significantly, there was little roof tile in amongst the stone tumble, suggesting that the roof had been removed, probably at the time that the doorway was blocked. Over the whole site only a fraction of the original roofing was recovered indicating that, along with the bulk of the stone, it had been taken for reuse elsewhere. Of the roofing material that was recovered, there was a wide variety of forms and fabrics suggesting that it too had originally been recycled from other buildings, some no doubt from earlier buildings on site, but also some from elsewhere. A rare find was a number of fragments of ridge tiles, in both a red and cream coloured fabric, some of which retained red paint suggesting that the white ones had had been painted to match the red. The floors consisted of a metalled surface, formed of small stones pressed into the clay. This floor survives in patches, and in between, a sub-layer of much larger ragstone pieces was revealed. The cella is subdivided into two chambers of roughly equal size, with a large double doorway between. There are also large double doorways between the corridor and the first cella chamber on the northern and eastern walls. In the inner chamber of the cella there is a small rectangular oven built into the north-western corner and in the outer chamber there is a large square oven, with a circular inset, that appears to be a bread oven, built into the south-western corner, although it is possible that this was constructed for a different purpose initially. These two oven structures, made of stone, appear to be part of the original structure of the building. Additionally, there are a number of more ad hoc, oven-like structures, that appear to date from late in the building’s life, inserted into this part of the building (see discussion of ovens below). A lead tablet, believed to be a defixio, was uncovered amongst the stone demolition layer and was found to contain a list of 14 names, both Celtic and Roman (Tomlin 2016, 415). Its identification as a defixio was due to the mirror writing that was employed. This practice was common on these ‘curse tablets’ and other examples are known from Bath and Uley (Gloucs.). A hoard of 154 copper alloy coins was found in the floor of the inner chamber. These coins are predominantly clipped copies and all date to the House of Constantine, ad 330-365. Several very worn coins of the House of Theodosius, ad 395-402, were found amongst the stone demolition (Holman 2018), consistent with similar finds from elsewhere on the site. These suggest a final demolition date at the end of the fourth century, or more likely given the worn condition of the coins, sometime in the fifth century. Building 5 appears to be a Romano-Celtic style temple, although its plan is unusual for Britain, with most parallels occurring on the Continent such as Cocquerel, Equevillon, and Kornelimunster. The nearest equivalent in Britain is perhaps Lamyatt Beacon (Somerset) or Dean Hall (Gloucs.). However, East Farleigh differs from these, as they are believed to be ‘podium’ temples, accessed via steps directly into the central cella section. Building 5 on the other hand is accessed via a doorway directly into the northern corridor. ROMAN BUILDINGS EXCAVATED BY MAAG AT EAST FARLEIGH: AN INTERIM REPORT 141 The main entrance faces the courtyard, to the north-east. This doorway was 2.39m wide and when discovered was blocked with mortared stone infill. The sides of the doorway were formed from semi-circular tufa blocks and suggested that the doorway was not furnished with doors and was designed to facilitate the passage of large numbers of people. There is a secondary external doorway of standard size on the eastern wall at the south-east corner, which would originally have been behind a wall connecting Building 5 to Building 3. This small single doorway in the south-eastern corner, is suggestive of an administrative access for temple officials, an impression that is supported by the joining wall between Buildings 3 and 5, which would have created an area at the rear not seen by anyone at the front of the building. This connecting wall was later removed sometime in the third or early fourth century. The internal, ‘temporary’ walls, may have created different ‘zones’ and may explain why there are two large doors into the cella, allowing the flow of visitors to be managed around the building. It may also be that there were other small shrines within the corridor; indeed, the end of the corridor at the southeast corner is the only part of the structure where decorative mouldings survive, indicating a special area which may have been a small wall shrine. A comparison maybe the octagonal temple at Nettleton Scrubb (Wilts.) which has an aedicula or shrine built into the inner wall of the ambulatory (Wedlake 1982, 44). At East Farleigh, the outer ambulatory wall of Building 5 does not survive beyond five or six courses above the Roman ground level, so it is impossible to tell for certain whether there were any windows. However, there were 47 pieces of window glass found at the site, predominantly from the vicinity of Building 5, suggesting that there were at least some windows in this building (Broadley 2019, 5). In parts of the building the internal walls were covered in plaster painted a dark red, which would have made the interior very dark, even on a bright day, and it would have been necessary to have used candles or lamps to navigate inside the building, especially if windows were absent in some areas. Internally the arrangement and decoration appear designed to enhance the mystical atmosphere of the buildings. What natural light there was would have come from clerestory windows high in the cella and would only have reached the ambulatory when the connecting doors were open. In the reconstruction of the buildings at East Farleigh the author has given them windows on the ambulatory simply because it seems sensible to have some natural light, but actually, there is very little evidence for windows in temple buildings. The nymphaeum at Gennes in France has windows in the apse, (Maligorne 2012, 137), and the huge cella at Autun in France also has windows on the ground floor (Lewis 1966, Plate 1a, 173), but neither of these examples are really comparable with the Romano-Celtic temples that are found in Britain. The only known site in Britain to have windows is Nettleton Scrubb, where the inner cella of the octagonal temple has a window, which the excavator believed was replicated on some of the other walls (which have not survived), allowing viewing from the ambulatory into the central area. Building 1: uncovered in 2005 this measures 27 by 14m and does not have any direct linkage to any of the other buildings, as far as we know. The level of preservation is poor, with only a single course of un-mortared stones being all that remains of most of the walls. An exception is the north-west corner where the remains of a STEPHEN CLIFTON 142 substantial mortared section of wall was revealed at a lower level than the rest of the building. This wall appears to be a surviving fragment of an earlier building, subsequently replaced by Building 1. Unfortunately, the significance of this was not appreciated at the time and it was not possible to explore it later. It is conceivable that the footings were retained because the walls were useful in this position to the replacement building, and the same technique can be seen with Buildings 2 and 3. Building 1 has three central rooms surrounded on three sides by a corridor. There only appears to be one cross wall interrupting the corridor (other than the possible earlier phase walls noted above), at the south-western corner. The ground plan of the building is strikingly similar to Building 5, albeit extended to accommodate an extra central room. Its orientation is in line with buildings 6 and 5, pointing north-east, although it is hard to be sure whether the main entranceway was on the eastern wall facing the open area, or on the southern wall as at Meonstoke (Hants.), which has a similar structure of comparable proportions. In the reconstruction of this building the author has placed the door facing the open area to the east, because, unlike the Meonstoke example, any door in the southern wall would have opened into the first chamber of the cella, rather than the corridor, as seems to be more common with these basilical buildings elsewhere. There is no direct evidence for any doors in any of the walls due to the level of demolition. This raises the question of what this building was used for. Was it another temple, only [this time] with a triple cella? Or was it some sort of meeting place? There do not appear to be any small rooms or cells which would normally be associated with sleeping quarters. There are other fairly similar examples, such as Keston, Thurnham and Darenth. The buildings at both Keston and Thurnham are similarly located and oriented in relation to the other associated buildings on the site. Building 6: is aligned to Buildings 1 and 5 and therefore seems to have been conceived as part of a group (Figs 11 and 12). The exact build sequence is hard to work out because all three blocks of buildings are independent of each other. However, the pottery evidence suggests a construction date at the end of the second or beginning of the third century (Lyne 2019). There were a number of coins found associated with the demolition layers, the latest of which was Theodosius I, ad 388-395 (Holman 2019). The building measures roughly 5.0 x 5.5m. The most striking thing about it is that there is a wall (0.88m wide), running through the middle of the building from south-west to north-east. This central wall had been demolished down to close to floor level within the building, effectively forming two ‘C’ shaped sections bisected by this central wall. There was no doorway between the two halves visible, although there was a small channel, 0.12m wide, that appeared to have been built into the structure, possibly to allow drainage from one chamber to the other. On the eastern and western side of the building were two openings, at 2.61m and 2.89m respectively (by comparison the external opening to Building 5 was 2.39m), which emphasises the different treatment of the two chambers and perhaps suggests that they were not originally intended to be seen simultaneously. These openings are not symmetrically in the centre of their respective walls, being closer to the southern ends. There is no evidence to suggest that they had doors. On both the south-western and south-eastern doorjambs there is evidence, in the ROMAN BUILDINGS EXCAVATED BY MAAG AT EAST FARLEIGH: AN INTERIM REPORT 143 Fig. 11 Building 6 showing development phases (Clifton 2017). STEPHEN CLIFTON 144 shape of a spread of mortar, that at some time the openings had been altered to become narrower. But at the time of excavation no more substantial evidence of this alteration survived, suggesting that if it had been altered in this way, it had subsequently reverted to its original size or was a very insubstantial alteration. On the inside of the southern wall of the building, where the central wall meets the outer walls there is a niche created from stone and mortar set into the wall. The niche is a small rectangular inset fashioned out of reddish orange mortar that could not have been constructed whilst the central wall was in place. (Similar features are seen at Dean Hall, Gloucs.) The wall’s removal would have turned the building into a possible gateway, given the two opposing entrances. However, there are no obvious signs of wear, or cart ruts, in the remains of the central wall to indicate traffic from one side to the other. If the central wall external to the building was removed at the same time, the land to the east and west of the building would have been accessible without the need to go through the building. In the eastern chamber two stone ‘benches’ were found. These could perhaps be interpreted as exedre, or cult benches, where worshippers could come and share a meal with the deity. The floor in both chambers appears to have been a metalled surface, with small pieces of stone pushed into the soil. However, this floor does not extend from east to west due to the intervening central wall which protrudes slightly above the floor level. The metalling on both sides of the building does appear to extend beyond the entrances and was followed for 1.5m to the east, suggesting a trackway or paved area Fig. 12 Building 6 (Clifton 2017). ROMAN BUILDINGS EXCAVATED BY MAAG AT EAST FARLEIGH: AN INTERIM REPORT 145 in front of the entrances. The central wall extends beyond the building to the north and south. Time constraints meant that the wall could not be followed more than a few metres either side. However, it was possible to determine that the wall ran 7.5m to the south, and at least 5.5m to the north, although not much more. No corners were observed. Nevertheless, it is possible to infer from where the wall was absent that it must have turned to the east at both ends, creating one end of an enclosure, with the building roughly central on its western wall. The land to the east of Building 6 was not available during the period of excavation to test this theory further. The next phase seems to be the removal of the central wall externally to the building. It is possible that this occurred simultaneously with the removal of the wall within the building. However, the utility of the building would have altered as it would then be freestanding. It may be that the narrowing of the entrances may have facilitated the fitting of doors to secure the building as a detached structure. The addition of a low narrow wall butted against the north-west corner of the building heading north-east towards the river was the next development. This wall was roughly 0.34m thick and constructed of dressed ragstone blocks back-to-back. The wall was visible as a single course over a 12m length, laid directly onto the subsoil with no foundation and no mortar, and was robbed out for the last 2m. It led to a metalled surface of which it was possible to excavate only a small area. A cemetery wall possibly; however, no burials were discovered, and it would have been customary for such a wall to completely enclose the space, and there is no evidence of that here. At some point before Building 6 was finally demolished it was apparently used for a different purpose. There was a feature cut into the floor of the building predominantly on the western side, but terminating on the eastern side, cutting through the central wall. This feature looks like a classic ‘corn-dryer’, however there is no sign of any burning at the stoke end and at the other end there is a hole, produced by removing the wall and floor of the building and some of the subfloor stone make-up material. When excavated this hole was largely filled with a substantial piece of ragstone which appeared to be a ritual act of cessation. This feature looks more like some sort of sluice, than a corn-dryer, but there were no other clues to shed light on its function. It was in turn covered by demolition material from the building. There are no obvious parallels for this building in Britain, although it does appear to be reminiscent of small shrines but here installed back-to-back associated with a possible temenos. The 1839 Building: after 12 years excavating at the site, MAAG had uncovered at least six Roman buildings but still there was no sign of the building that was found in 1839 (Fig. 1). It is likely that much of this building was removed to make way for the construction of hop-picker’s accommodation, and associated tracks and revetments in the 1830s. The author has speculatively included it in the site reconstruction based on the antiquarian references and deduction, although there is no direct archaeological evidence for it as yet. It is not known which phase it would have belonged to but as it has much the same orientation as the other laterphase buildings, it has been assumed that it may have been contemporary with them and survived into the fourth century. From Smith’s plan it appears to be an aisled building, at least 18.5m long and STEPHEN CLIFTON 146 13.4m wide. The full ground plan is not recorded, probably because only the area that they were intending to clear was exposed at the time. No indication of wall thickness is given, but the overall proportions look similar to Building 1. The main differences between them are the internal wall arrangements. This building was almost certainly much longer than shown on the partial plan, and there are walls continuing to the north-east. This building has parallels at many other sites in Kent and appears to be quite a common, if adaptable form. Already mentioned is the similarity with the basilical building at Meonstoke, but another building at Hollingbourne, which was located primarily through resistivity survey work, also appears to be about the same proportions and orientation and is similarly sited on the bank of a watercourse (Feakes 2008, 19). The Ditches: a series of ditches that appear to have been used for draining water away from the buildings terminate just in front of the eastern end of Building 3. Excavation revealed the primary system running off to the east with a tributary joining from the south, down the hill. A magnetometry survey of the adjacent parcel of land to the east showed the ditch running along the line of slope, parallel to the river for a further 100m (Taylor 2017). This would suggest that they were designed to divert water running down the hill into these water channels away from as yet undetected buildings. It may also be that the channels, once full of water, may have been deliberately created for other purposes, possibly ritual. Another possibility is that the ditch is taking water from an undiscovered bathhouse up the slope to the south. This may seem fanciful, but it has been a suggestion owing to the presence of box flue tiles on the site, but no hypocausts in any of the excavated buildings. In the space between Buildings 3 and 4 evidence of construction work was found. There were several phases of mortar mixing and deposits of crushed tile or baked clay. There was also evidence for a channel, timber-lined close to the buildings, which was traced for a further 35m in a gulley before it was truncated by the revetment. This may have been for taking water away, avoiding the nearby structures. The final years at East Farleigh Sometime towards the end of the third century the site appears to have gone into decline. Pottery evidence would suggest that Building 5 was still standing in the mid fourth century and this was when the north-eastern doorway was walled up (Lyne 2019). However, activity at the building appears to continue, with at least five ovens being inserted, and quern stones and mortaria left broken inside. The outer wall of the ambulatory was removed from the north-eastern corner and some of the cella at the same point, presumably to facilitate entry to this area. Late fourth-century coins from the demolition layer over Buildings 5 and 6 suggest that these were finally demolished at the end of the fourth century or the early years of the fifth (Holman 2019). However, unlike Buildings 1, 6 and the 1839 one, the cella of Building 5, and at least some of Building 3 remained standing until this final demise, whether by natural collapse or purposeful demolition. The other buildings were dismantled down to ground level in most places, and the building ROMAN BUILDINGS EXCAVATED BY MAAG AT EAST FARLEIGH: AN INTERIM REPORT 147 materials apparently removed from the site. Building 6 clearly has evidence for later re-use at the end of the fourth century, with the insertion of the sluice, and the later phase narrow wall abutted against the north-west corner. The question therefore arises, who was reusing these buildings at the end of their life, and what for? If it is accepted that Building 5 was walled up in the mid fourth century, then there was half a century of potential use after this. Perhaps the ‘landowner’ sealed off these buildings, possibly simultaneously removing the tiles from the roof, so that the building could not be used. But the local populace may have still regarded it as a sacred site and used it as such. This later informal use is noted by Farquhar at the Progress Romano-British site near Otford (Farquhar 2017, 23), and other examples include Nettleton Scrubb (Wedlake 1982, 81), where a late shrine was constructed from building debris. At East Farleigh there are also several infant burials close to some of the standing walls, one of which was inside the south-east corner of Building 5 on the floor surface and covered with crushed burnt clay in the area where moulded cornicing may have indicated a shrine. The ovens are a curious feature and have caused much discussion with it being suggested that the building was reused as a kitchen (Daniels 2010, 12), but another explanation may be the continued use of these as part of the rituals, either for cooking sacrificial animal parts, or for baking bread as an offering or part of the ritual. Also found were a number of broken quern stones and mortaria. These have been found at other sacred sites and quern stones are often found in structured deposits, frequently deliberately broken. It is unlikely that anyone was living in the ruined buildings; however, there are a number of post-holes from around the buildings which may indicate additional shelter was built, possibly on a temporary basis. Unfortunately, these post-holes contain no dating evidence, and stratigraphically they appear sometime between the backfilling of the Iron Age ditches, approximately ad 150/170, and the demolition of the buildings, c.400. It maybe that we are seeing a seasonal return to the site, perhaps to celebrate religious festivals by those still adhering to the deities or cults at the site, presumably without the knowledge or permission of the ‘landowner’, if indeed there was one by this time. If there was a river crossing, as seems likely, then it is possible that this was still in use even though the religious facilities may have been in ruins. Sacred Attribution versus Villa There are a number of sites described as ‘villas’, such as East Farleigh, that do not make sense as such and in fact have a number of characteristics which mark them out as something different. Often it is the lack of features that should make us question our assumptions. In the case of East Farleigh, the unusual arrangement of buildings and the lack of any clearly ‘domestic’ facilities was the first indication that we were not dealing with a villa. The orientation to the north-east, when most villas in south-east Britain face broadly to the south, was another. The lack of any obvious bathhouse or any hypocausted rooms was also a factor. The proximity of the East Farleigh buildings to the possible Barming villa is difficult to appreciate due to the modern plantation of trees obscuring the view, but to have two villas so close, albeit divided by river, seems unlikely. The underlying Iron Age activity on the site is a clue to the nature of the later STEPHEN CLIFTON 148 Roman-period buildings located here. If we are seeing an earlier ritual landscape then the later buildings are merely a monumentalisation of the pre-existing arrangement. This is supported by Casey’s ‘development economics’, where new wealth generated by local entrepreneurs is invested in new sacred facilities on traditional sites and at key junctions and river crossings (Casey 1992, 98). In many ways, the rise of the rural temple can be seen as part of the villa phenomenon and goes hand-inhand with the growing prosperity of the province (Smith et al. 2018, 135). This is reflected in the spread of both Romano-Celtic temples and villas which predominate in the more affluent south-east (ibid., 133). Villa estate boundaries were indicated by property markers which had to be consecrated with a sacrifice, and villas without temple facilities were seen as ‘unprotected’ and were expected to wither and not prosper (Bowes 2006, 74). There are many examples of temples associated with villas, referred to as ‘Villenheiligtümer’ (estate shrines), such as Otrang and Newel in Gallia Belgica and Darenth and Bancroft (Bucks.) in Britain, but some sites of religious activity are close to a villa, but clearly separate, as at East Farleigh. These sites seem to be something slightly different. They are well-located, yet possibly part of an estate, placed to entice travellers and pilgrims alike to the facility in order to profit from the sale of votive items and hospitality (Woodward 1992, 47; Aldhouse- Green 2018, 88). Many were located on pre-Roman sacred sites. At many sacred sites there are numerous buildings that do not appear to be temples or shrines, and which are difficult to identify, such as at Uley in Gloucestershire and Nettleton Scrubb (Woodward 1992, 49; Smith 2000a, 38). As King points out, ‘it is clear that the architecture of ancillary buildings at temple sites can often resemble villa buildings and can be confused with them’ (King 2018, 9). Many of the conclusions in King’s paper on the site at Meonstoke echo the thinking that took place at East Farleigh (ibid., 5). However, there has perhaps been a tendency to over-categorise, suggesting a site is either a villa or a religious sanctuary (Bowes 2006, 73), and this may be a reflection of modern prejudices rather than a realisation of the true nature of the Roman landscape. It is hard for us now to appreciate the extent to which religion permeated every facet of life in the Roman world. Almost every action in daily life might have been accompanied by ritual or reference to an appropriate deity. Roads and river crossings would have been prime spots for a consultation to ensure that the omens were good (Derks 1998, 144), and consequently many temples are found close to roads and on the banks of rivers. The buildings at East Farleigh being one example but sites such as Nettleton Scrubb and Springhead in Kent are similarly sited. The Finds Assemblage MAAG’s work at East Farleigh was blessed with a wealth of small finds including over 300 coins among which were found a hoard and a very rare silver minim inscribed ‘SEGO’ and believed to be from a Kentish Iron Age king, dated ad 1-10 (Holman 2019). Also recovered were 25 boxes of animal bone (yet to be finally evaluated), and a great deal of building material, including roof tiles, painted wall plaster and plaster mouldings. There is not room in this summary to explore this in detail, but the pattern of finds has a bearing on the suggested sacred nature of the site, yet in and of themselves they do not identify a site as sacred, many of them ROMAN BUILDINGS EXCAVATED BY MAAG AT EAST FARLEIGH: AN INTERIM REPORT 149 being perfectly ordinary everyday items that might be expected to be recovered from a domestic setting. Indeed, it is this mundane aspect that has often contributed to the difficulty in separating secular from sacred. On top of this, the ubiquity of ritual practice in the Roman world has made it possible for either interpretation to be possible, or indeed a mixture of both (Henig 1984, 128), and it is often the subjective opinion of the excavator as to the attribution a particular find or site is given, which can result in years of misattribution, such as at Meonstoke (King 2018, 5). The assemblage of finds at East Farleigh has many parallels with other sacred sites and is equally unremarkable in not providing a pointer towards a particular deity or ritual activity. Although there is one building believed to be a temple, there are several other buildings which could also have been shrines dedicated to other gods. The nature of the sacred beliefs and the identity of the particular deity have often been the subject of speculation based on the ‘votive’ objects. However, it may just be that people merely used whatever they owned that looked like a valuable object, irrespective of the nature of the deity. It is also suggested that items could be purchased at the temple for just this purpose. The ‘hoard’, found in the sub-floor layer of the cella of Building 5 consisted almost entirely of coins that were clipped; this is very common at cult sites and perhaps points to their preparation specifically for votive activity (Woodward 1992, 67). There were also a number of pieces of copper and lead sheet, which were uninscribed in any way, but which would most likely have been shiny when deposited. There is no indication of any sort of shop or kiosk amongst the buildings excavated so far, however there are similarities in the rings and bracelets found, which might indicate a common source produced at or near the complex. The lead defixio found at East Farleigh contains a list of 14 names (Fig. 13), some using ‘mirror writing’ to add extra potency but alas gives no further clue to the circumstances of its deposition or the deity invoked (Tomlin 2016, 415). At East Farleigh many of the objects have been broken, often in more than one place, indicating a deliberate act. A good example is a bone stylus, broken in two places, or a mysterious jet object, broken at both ends. This is often described as the ritual ‘killing’ of an object by bending, breaking or disfiguring it. There are a number of categories of finds from East Farleigh that accord with assemblages found at other known sacred sites; a collection of 36 antler tools and horn cores, including one antler base shaped to form an amulet; 17 snapped fragments of copper alloy bracelets; 15 copper alloy rings including a key ring; nine pieces of lead or copper sheets including the defixio; two fragments of pipeclay ‘Venus’ figurines. Also found at East Farleigh were a number of fossils including a small fossilised sea urchin. Seen as curiosities when discovered, they are echoed by similar finds at Meonstoke and Farley Heath (Surrey) suggesting that these were brought to temples as offerings due to their unusual nature (King 2018, 6). At many Roman period sites there is little left of the building other than the footprint of the walls, and it is difficult to project a three-dimensional structure from a two-dimensional ground plan. It is often the pattern of finds that confirms the site as sacred, such as at Kelvedon (Essex) where the assemblage was key to its identification as a sacred site (Smith 2000b, 98). It is not always straightforward though, as we see a bewildering array of material goods, supposedly lost or discarded at almost all Romano-British sites in Britain, especially villa sites, making STEPHEN CLIFTON 150 Fig. 13 Defixio found at East Farleigh (illustration from R.S.O. Tomlin 2016). ROMAN BUILDINGS EXCAVATED BY MAAG AT EAST FARLEIGH: AN INTERIM REPORT 151 the distinction between domestic refuse and ritual offering especially finely balanced (ibid., 59). Orientation and Location At East Farleigh the principal entranceway to the temple (Building 5) is very conspicuous and the buildings clearly faced north-east, although due to the nature of the site around an open courtyard area, some of the doorways will inevitably face in other directions, such as those of Building 6 which has one entrance to the northwest and the other to the south-east. It is not necessarily that the every individual building is orientated to the north-east, but that the complex as a whole is angled in this direction. It may be significant that it is only at this angle that the sun never penetrates directly, and buildings can remain in permanent shade, especially if there are very few windows, which may have manifested as a sacred preference. The proximity of a river or lesser watercourse, often to the north of the site, is a recurring theme of many sacred sites. This is likely to be accompanied by the presence of a crossing and associated roads. This is certainly true of East Farleigh, which sits on the south bank of the river Medway, and indeed may have been close to springs that issue from the ragstone bedrock into the river from time to time. Smith says that a third of all temple sites in Britain are on or near rivers (Smith 2000a, 312). Springs are obviously another focus, with Bath and Springhead as prominent examples. Sites on the coast, often near estuaries, are also a common location, such as Minster in Thanet (Parfitt 2006). Casey suggests that the answer lies in the arrival of newfound wealth being invested in new facilities on ancestral land, partly out of religiosity and munificence, but also as a visible manifestation of that wealth and as a sound investment (Casey 1992, 99). Ovens and Temples At East Farleigh there appear to be at least four phases of ovens. Firstly, there are the small oval-shaped structures dug into the floor of Building 3. These may predate the construction of the other buildings (5, 1 and 6), and more than likely carried on in use in support of the other structures, eventually being superseded by the corn-dryer type ovens seen at the western end of the building and finally the one in the middle of the structure. It is unclear what these were used for, but presumably some sort of baking connected with the rituals, or perhaps just to feed the assembled worshippers. Then there are the other oven-like structures built into the fabric of Building 5 (Fig. 14). There may have been others elsewhere on the site, but they have not survived. The two inside the building, both constructed of mortared stone and built into a corner, are very different to each other. The larger of the two (the ‘breadoven’) in the outer chamber was clearly contemporary with the construction of the building, as evidenced by the offsetting of the doors through to the inner chamber, if the doors had been central, they would have fouled the oven and it would have been too hot and cramped. It is conceivable that the structure was not originally intended as an oven; an alternative interpretation as the base for a lead tank or cult statue is possible. Any daub superstructure and plasterwork has disappeared, so it STEPHEN CLIFTON 152 Fig. 14 Building 5: (a) outer cella room showing various oven structures; (b) showing oven built into corner of the inner cella room. (Clifton 2010). a b ROMAN BUILDINGS EXCAVATED BY MAAG AT EAST FARLEIGH: AN INTERIM REPORT 153 is impossible to say for sure, but the presence of reddened clay and stone within the circular structure makes its use as a hearth or oven seem likely, at least periodically. The other oven structure in the inner chamber is rectangular and much smaller and was found choked up with red and black ash and soot. The obvious question is why there are two such structures within the building, albeit of different proportions, which clearly had different functions. In the later phase of the building’s life, after the main entrance has been walled up, and subsequently demolished, a number of other, more ad hoc, oven structures were dug into the two central chambers. In the outer room a corn-dryer with a crooked flue was built, and then a secondary structure built over the top once it had become choked up. There was also a small wattle and daub oven with a roof tile as a base, built into the middle of the floor. At least two other similar structures were inserted into the inner room, along with quern stones and two pots dug into the floor. The sheer number and variety of these structures suggests use over a long time, rather than some sort of industrial use, and supports the idea of people returning to celebrate festivals at regular times of the year and continuing to carry out the appropriate rituals which presumably involved their use. Ovens at temple sites are quite common, and are perhaps one of the various foci of temples, alongside statues, altars and pools (Smith 2000a, 319). Ovens have been found at 17 sites in Britain, such as Verulamium’s ‘triangular temple’, Titsey in Surrey, Chanctonbury Ring in Sussex, Newington and Springhead (temple 1) in Kent. Most ovens were external to the temple structure, and this may be to avoid the risk of fire, or it may be because the associated activities took place outside the building. At only two sites (other than East Farleigh), Springhead and Brigstock, Northants., are the ovens actually inside the main temple structures. What we do not know is how exactly they were used. It has been suggested that the ovens were for preparing the sacrificed animals prior to the feast at the end of the ceremony (Greenfield 1963, 68). The examples from East Farleigh are of various sizes and shapes and clearly were intended for different purposes. Many of the structures would have been too small to have been used to cook the carcass of sheep or goat. Even if the animals had been butchered, the ovens were unsuited for use as a barbecue. They are much more reminiscent of bread-making facilities. A Kentish parallel would be the site of Roman buildings in Broadstairs where numerous ovens had been dug through the floor of a rectangular building displacing several infant burials (Moody 2007, 202); this building has the size and proportions of a shrine or small temple, and the ovens clearly demonstrated continued re-use over a long period. Smith suggests that the frequency of ovens on sacred sites points to their importance within the ceremonies, and that the occurrence of mortaria are to be expected as part of the food preparation associated with the rituals (Smith 2000a, 330). Broken quern stones are another common find, and have usually been assumed to be votive offerings, rather than part of the ritual (ibid., 319). The ovens at East Farleigh were accompanied by both quern stones and mortaria fragments within the buildings. Presumably smoke from internal ovens was not an issue, merely adding to the mystical atmosphere before making its way out through the tiles on the roof, thereby confirming the lack of a ceiling and upper storey. It may also suggest that the ovens were fired up under very supervised and perhaps infrequent occasions to minimise any risk of fire. STEPHEN CLIFTON 154 Fig. 15 Drawing of the complex from the north-east (Clifton 20 20). ROMAN BUILDINGS EXCAVATED BY MAAG AT EAST FARLEIGH: AN INTERIM REPORT 155 Conclusions and final thoughts on East Farleigh MAAG have been able to establish that there was an Iron Age presence on the site before the buildings were constructed, possibly an oppidum. There are at least three distinct phases of buildings on the site, it finally being abandoned in the late fourth or early fifth century. The last phase of buildings was constructed in the early third century, around a courtyard area oriented north-east towards the river Medway and a possible villa on the far bank at Barming. Of these buildings at least one was a temple, to an unknown deity (or deities). Two others are possible religious buildings. (Fig. 15 shows an overall view of the complex in its final phase.) It is not clear whether the earlier phases were also of a religious nature. There is no evidence that any of the buildings are associated with a potential, as yet undiscovered, villa on the south side of the river, although it is possible that the buildings on the north side may have been part of the same Barming estate with a river crossing and associated roads. The buildings that were found in the nineteenth century were not identified during the recent programme of excavations. However, it is likely, based on the description that Smith gives us, that they were removed during the construction of the hoppicker’s accommodation and the associated farm tracks. It has been remarked by visitors to the site how well constructed the buildings were. One observer compared them to some of the mile castles on Hadrian’s wall. Whilst there is no suggestion of military involvement, it is clear that the buildings were very competently constructed and would have represented a substantial financial investment. We know that at least one of the buildings was plastered on the outside, yet possibly only on one wall, as no other wall plaster was found in association with the other outer walls. This might indicate that this finish was to be employed on other walls and buildings, but for some reason was not completed. Examples of the roof tiles were retained during the excavation and form only a representative sample from the many fragments encountered. But the whole assemblage would only have been a fraction of what would have been needed to roof all of the buildings, suggesting that the vast majority had been removed to be reused elsewhere. By the fourth century any new building would have had to rely on recycled building material. In fact, the huge variety of different fabrics, thickness and colour of the retained examples of roof tile suggest that they in turn had been recycled from other buildings in the third century. Also recovered were a good selection of box flue tiles, yet none of the buildings were equipped with hypocausts. During excavation it was speculated that somewhere there must be a bathhouse or a hypocausted building, yet none materialised. Another explanation might be that they arrived at the site as part of the demolished remains of another building, despite not being needed. Although MAAG excavated the site for twelve seasons, it was not possible to explore everything, and it is fair to say that there are undoubtedly many more years of archaeological work that could be done on the site to further support, or dispute, the conclusions proposed here. MAAG has continued to work on land close to the buildings described here, where Iron Age and sub-Roman features in the wider landscape are adding to the picture that we already have of Romano-British East Farleigh. The site at East Farleigh is remarkable for its level of preservation, and deserves to be appreciated for its importance both within Kent, and nationally. STEPHEN CLIFTON 156 acknowledgements This project could not have been completed without the unique character of Albert Daniels who did so much to get us to where we are, and who’s recent loss is deeply felt; thanks are due to Keith Parfitt, David Holman, Malcolm Lyne, Rose Broadley, Ges Moody, Richard Taylor, Andrew Mayfield, Paul Wilkinson and Steve Willis, all of whom contributed expert advice and encouragement; thanks are also due to KAS for a grant to fund specialist reports; to Gallaghers Cement; huge thanks are also due to the generosity of the late Bryan Boughan and Jo Boughan for allowing us to dig holes in their property for so long unmolested; and lastly the hours of work that all of the MAAG volunteers put in must be recognised and applauded. bibliography Aldhouse‐Green, M., 2018, Sacred Britannia – The Gods and Rituals of Roman Britain, Thames and Hudson Ltd, London. Andrews, C., 2001, ‘Romanisation: a Kentish perspective, Archaeologia Cantiana, 121, 25-42. Black, E., 2008, ‘Pagan religion in rural south-east Britain: contexts, deities and belief’, Ritual Landscapes of Roman South-East Britain, Oxford, Heritage Marketing/Oxbow Books, 1-25. Bowes, K., 2006, ‘Building sacred landscapes: villas and cult’, A. Chavarría Arnau, J. Arce and G.-P. Brogiolo (eds), Villas Tardoantiguas en el Mediterráneo Occidental (Anejos de AEspA XXXIX), Madrid, 73-95. Broadley, R., 2019, ‘Roman Glass from East Farleigh’, unpubl. report. Casey, P.J., 1992, ‘The monetization of a third world economy: Money supply in Britain in the first century AD’, in Current Research on the Romanisation of the Western Provinces, eds. M. Wood and F. Queiroga, BAR No. 575, 95-99. Clifton, S., 2019, ‘Excavations at East Farleigh 2019’, KAS Newsletter, 112, 6-8. Daniels, A.J., 2010, ‘East Farleigh Roman Buildings – August 2010 Update’, KAS Newsletter, 86, 12-13. Derks, T., 1998, Gods, temples, and ritual practices: the transformation of religious ideas and values in Roman Gaul, Amsterdam University Press. Detsicas, A., 1983, The Cantiaci, Alan Sutton Publishing, Gloucester. Farquhar, C., 2017, ‘An examination of the religious and ritual practices of the communities at the head of the Darent valley, Kent, in the Roman period’, unpubl. m.a. dissertation. Feakes, L., 2008, ‘Hollingbourne, Roman Finds’, in Discovering Ancient Lenham, vol. 1, Edition 2, 18-20. Greenfield, E., 1963, ‘The Romano-British shrines at Brigstock, Northants’, Antiquaries Journal, XLIII, 228-268. Henig, M., 1984, Religion in Roman Britain, Batsford, London. Holman, D., 2019, ‘East Farleigh Coin Report’ (unpubl.). Houliston, M., 1999, ‘Excavations at the Mount Roman villa, Maidstone, 1994’, Archaeologia Cantiana, 119, 71-172. King, A.C., 2018, ‘Excavations at Meonstoke Roman Site 2016 and 2017’, Hampshire Field Club and Archaeological Society Newsletter, 69. Lewis, M.J.T., 1966, Temples in Roman Britain, CUP. Lyne, M., 2019, ‘East Farleigh Pottery Report’ (unpubl.) Maligorne, Y., 2012, ‘La parure monumentale des agglomérations du territoire dans les cités de l’Ouest’, Aremorica, 5, 117-144. Margary, I.D., 1946, ‘Roman roads in west Kent’, Archaeologia Cantiana, 59, 29-63. ROMAN BUILDINGS EXCAVATED BY MAAG AT EAST FARLEIGH: AN INTERIM REPORT 157 Moody, G., 2007, ‘Iron Age and Romano-British Settlement at Bishop’s Avenue, North Foreland, Broadstairs’, Archaeologia Cantiana, 127, 197-211. Parfitt, K., 2006, ’The Roman Villa at Minster-in-Thanet. Part 3: The Corridor House, Building 4’, Archaeologia Cantiana, 135, 115-135. Payne, G., 1880, ‘The Remains of Roman Buildings Discovered at Barming near Maidstone, Archaeologia Cantiana, 13, 169 -170. Smith, A., 2000a, ‘The differential use of constructed sacred space in southern Britain, from the late Iron Age to the 4th century AD’, Vol. 1 (Doctoral dissertation, University of Glamorgan). Smith, A., 2000b, ditto Vol. 2. Smith, A., Allen, M., Brindle, T., Fulford, M., Lodwick, L. and Rohnbogner, A., 2018, Life and Death in the Countryside of Roman Britain, Britannia Monograph, 31, 120-204. Smith, J., 1839, Topography of Maidstone and its Environs, Printed and published by J. Smith, pp. 56-57. Taylor, J., 2007, An atlas of Roman rural settlement in England Council for British Archaeology, vol. 151. Tomlin, R.S.O., 2016, ‘Inscriptions’, Britannia, Vol. 47, 414-415. Wedlake, W.J., 1982, The excavation of the Shrine of Apollo at Nettleton, Wiltshire, 1956- 1971. Woodward, A., 1992, Shrines and Sacrifice, English Heritage, BT Batsford Ltd.