The Roman Villa at Minster-in-Thanet

333 THE ROMAN VILLA AT MINSTER-IN-THANET. PART 6: THE VILLA ENCLOSURE; BUILDINGS 2 AND 5 keith parfitt, emma boast and gerald moody This is the sixth report detailing results of the Society’s excavations, undertaken jointly with the Trust for Thanet Archaeology and the Thanet Archaeological Society, at the Roman villa site on Abbey Farm, Minster-in-Thanet (NGR TR 3135 6463, centred). It describes the boundary walls enclosing the main villa house (Building 1), Building 2 and Building 5 (minor structures on the periphery of the complex), together with several ditches and some of the other miscellaneous Roman features that lay outside the main enclosure. Building 5 and Wall Foundation, 9039 Along the northern side of the villa enclosure a number of foundations, distinctively constructed from small flint pebbles, was located. These appeared to pre-date the chalk footings of the main villa boundary walls and included the remains of a small building (Figs 1 and 2, B5), together with the foundation for a substantial length of e-w boundary wall (Figs 1 and 3, 9039), set on a different axis to the north wall of the later villa enclosure. The position of Building 5 seems to have had some influence on the siting of the subsequent boundary walls because it formed the north-east corner of the villa enclosure. Whether this small pre-existing structure had continued as a functioning building or simply provided a conveniently located shell whose walls were partially re-used, could not be determined from the surviving remains. Building 5 A large area cleared at the north-east corner of the villa enclosure in 2002 somewhat unexpectedly revealed the pebble footings of a small rectangular building. This was designated Building 5 (Figs 1 and 2, B5). From the surviving remains, a structure with internal dimensions of about KEITH PARFITT, EMMA BOAST ANDand gerald moody 334 Fig. 1 General plan of the villa complex showing the position of excavated buildings in relation to the enclosure wall, outer ditches and surface contours. ROMAN VILLA, MINSTER-IN-THANET. PART VI: ENCLOSURE, BUILDINGS 2 & 5 335 Fig. 2 General plan of the NE area showing Building 5, Shaft F. 821 and Ditch F. 804. KEITH PARFITT, EMMA BOAST ANDand gerald moody 336 4.00m (n-s) by at least 2.20m (e-w) was indicated. Its eastern foundation survived complete but only short sections of the north and south wall foundations remained. There was no trace of any west foundation which, if it ever existed, may have been destroyed after the structure had been incorporated into the main villa enclosure (see below). The remaining footings were about 0.65m wide and 0.18m deep, mainly composed of small black flint pebbles with some small chalk lumps mixed with a few oyster shells, set in brown clay loam. The chalk foundation of the east villa enclosure wall clearly butted onto the south-east corner of Building 5 demonstrating that it was already in existence (Fig. 2). Less conclusively, on the north side the villa boundary wall stopped about 2.50m short of the surviving footing of the structure. Boundary Wall Foundation 9039 West of Building 5, but falling on the same line as its south wall, another long section of pebble foundation (9039) was located. The course of this footing fell just inside the north boundary wall of the main villa enclosure but lay on a slightly different axis, making it clear that the two walls represented could not have been contemporary (Figs 1 and 3). The pebble foundation was traced for minimum distance of about 32m. It was again constructed of small black pebbles, set in orange-brown sandy loam with chalk specks. The eastern end of the foundation was located about 20m west of Building 5 (Fig. 1). There was no evidence for any return here, indicating that the footing represented a straight boundary wall rather than forming part of another building. The western end of the foundation lay beyond the excavation limits but it appeared to have terminated before reaching the north-west corner of the villa enclosure (Fig. 3). Following the demolition of the associated wall, the remaining foundation at the east end had been sealed by a thin layer of rough pebble metalling (4022, not shown on plan). The make-up of this metalling produced a few fragments of Roman tile and two first/second-century pot-sherds. The construction, together with their relative positioning, suggests that these pebble foundations are contemporary. Whether they relate to structures connected with the earliest phases of occupation at the main villa or are associated with an earlier structural complex remains uncertain. However, if the main walled enclosure is contemporary with the earliest phases of the villa house, as is suggested below (see Discussion below), these pebble foundations must pre-date the villa complex. The Main Villa Enclosure The principal house of the villa complex (Building 1) and its adjacent ROMAN VILLA, MINSTER-IN-THANET. PART VI: ENCLOSURE, BUILDINGS 2 & 5 337 Fig. 3 General plan of the centre-north area, showing details of boundary walls, pits, ditches and fence line. KEITH PARFITT, EMMA BOAST ANDand gerald moody 338 bath-suite (Building 3), both previously reported (Perkins 2004; Parfitt et al. 2008), were set inside a rectangular walled enclosure (Fig. 1). Further downhill, adjoining the south-east and south-west corners of this enclosure, Buildings 4 and 6 had subsequently been butted onto the outside of the enclosure’s south wall (Parfitt 2006, 117; 2007, 267). Overall, the area delimited by the walls was not quite an exact rectangle and measured internally 67 to 72m (e-w) by 117 to 121m (n-s). Thus, the walls enclosed an area of around 8,400m2 (0.84ha, or 2 acres) but only their shallow foundations had survived. During the course of the excavations the shorter north and south boundary walls of the enclosure were mostly exposed, whilst significant portions of the longer east and west walls were revealed in a series of trenches and area excavations (Fig. 1). The corners of the enclosure at the north-east, north-west and south-west were examined but the south-east corner, in the area of Building 4, was found to have been completely destroyed by the subsequent expansion of that building and by later plough erosion (Parfitt 2006, fig. 2). The north-east corner had apparently been formed by incorporating the walls of the earlier Building 5 (see above). Everywhere damaged by the plough, the surviving foundations for the villa enclosure wall consisted of a shallow footing of rammed chalk rubble, with some small flint pebbles and occasional larger flints, set in a flat-bottomed construction trench. The better preserved sections of this foundation varied between 0.60 and 0.80m in width and were up to 0.28m deep. Traces of contemporary buttresses projecting from the boundary walls were recorded at various points (Fig. 1, marked ‘b’). Most of these were external but two internal ones were noted along the west wall, together with one on the north wall and another on the east wall (see below). The buttresses did not appear to have been evenly spaced but several only survived as eroded stubs and an unknown number had probably been completely destroyed through plough erosion. There was evidence for a gateway centrally positioned in the south boundary wall (detailed below). This is likely to have been the principal entrance into the enclosure as the entire villa complex seems to have been laid out facing downhill to the south, apparently intended to impress visitors approaching from this direction (Parfitt 2007, 294). Further gateways very probably existed in the other boundary walls but no clear evidence for their positions could be identified. The East Boundary Wall The line of the east boundary wall was established by excavations at the north-east and south-east corners of the enclosure, supplemented by a ROMAN VILLA, MINSTER-IN-THANET. PART VI: ENCLOSURE, BUILDINGS 2 & 5 339 number of intermediate trenches (Fig. 1). The wall had a total surviving length of some 105m but must have originally been about 116m. Preservation at the southern end was poor and an 11m length, including all evidence for the south-east corner itself, had been completely destroyed in the area of Building 4 (Parfitt 2006, fig. 2). The position of two buttresses projecting from the main foundation, one internal and one external, were identified. The internal buttress lay some 3m south of Building 5; the external one was positioned about 16m south of this. The South Boundary Wall and Gateway Almost the complete surviving length of the south boundary wall foundation was exposed (Parfitt 2006, fig. 2; Parfitt 2007, fig. 2; Figs 4 and 5, 7207). It was about 69m in total surviving length and was provided with at least seven externally projecting buttresses (Figs 1 and 4). Near to the corners, Buildings 4 and 6 had been subsequently butted onto the outside of this wall (Parfitt 2006, 117; 2007, 267). Although there was good evidence for a centrally positioned gateway through this boundary wall in later phases, the size and form of any original entrance here was less clear. There was no indication that there was ever a break in the wall foundation and no remains of any special structures, such as large gate-piers or a gate-house, were revealed. It may be reasonably assumed that the earliest entrance had been set in the 5m long section of wall that lay between the two central buttresses located. Whatever the original arrangements, there was much clearer evidence for a gateway in this area during the later phases of the villa’s use. The foundations of the original enclosure wall had subsequently been cut through by two massive post-pits (Fig. 4, Fs 7219 and 7221), each over 1.00m across and about 1.40m deep. Pit F. 7219 had destroyed most of the original centre-west buttress suggesting that there might have been a slight westerly shift in the positioning of the new entrance in relation to the earlier one. There can be no doubt that these post-pits once held a pair of substantial timber gate-posts. The outline of an oval post-pipe survived within each pit (Fig. 5). These were between 0.30 and 0.60m across and were set some 3.35m apart, centre to centre. This suggests that the gateway was about 3m wide. A shallow slot (Fig. 4, F. 7210) running between the two posts appeared to represent the position of a timber sill beam/door-stop. Traces of a rough pebble metalling (Fig. 4, 7206) extended away to the north and south of the gateway and this must represent an access roadway passing through the entrance. Partially sealing the original wall foundation between the gate-posts, the surface of this metalling north of the gateway was slightly cambered. There were no neat edges to the metalling and it may have been largely confined to the area of the gateway. Certainly, no KEITH PARFITT, EMMA BOAST ANDand gerald moody 340 Fig. 4 General plan of the south gate area. ROMAN VILLA, MINSTER-IN-THANET. PART VI: ENCLOSURE, BUILDINGS 2 & 5 341 Fig. 5 Sections across gate post-pits, Fs 7219 and 7221. KEITH PARFITT, EMMA BOAST ANDand gerald moody 342 traces of it were located in trenching across the villa enclosure further to the north or to the south, outside the enclosure. If it ever existed, it had most probably been destroyed by subsequent ploughing in these areas. In the actual gateway itself the metalling had been partially worn away to re-expose the original chalk wall foundation. The surface of the foundation here was noticeably worn and clearly it had been regularly passed over by traffic. Five residual pot-sherds of first/second-century date were found in the metalling but of rather more significance was the single Roman coin it also contained. This was found adjacent to the east gate-post (Fig. 4) and has been identified as a plated, hybrid copy of a denarius of Septimus Severus, dating to ad 200-209 (Holman and Parfitt 2005, Coin 2002/26). Its discovery indicates that the metalling here cannot have been laid before the early third century, although whether the coin was contained within a late repair, rather than the primary surfacing, remains uncertain. Eight more pot-sherds came from the filling of the main post-pits. These range in date from mid-first to the mid-third centuries. There was evidence for later changes to the gateway. The metalling and the filling of the two main post-pits had subsequently been cut into by a number of smaller post-holes (Fig. 4, Fs 7222, 7224, 7255, 7257, 7259, 7261, 7263 and 7265). These are unlikely to all be of the same date. Some (e.g. Fs 7222, 7224 and 7265) could represent various replacement gate-posts, whilst a setting of four, Fs 7257, 7259, 7261 and 7263, cut into the metalling seem to be contemporary and specifically connected with defining a much smaller entrance, less than 1.00m across. No useful dating evidence was recovered from any of these later post-holes but they must relate to the final phases of use of the villa enclosure. The West Boundary Wall The line of this wall was established by the location of the north-west and south-west corners of the enclosure, supplemented by the excavation of a series of intermediate trenches (Fig. 1). It had a total length of about 118m. Substantial sections of the foundation were exposed towards the northern end during the investigation of Buildings 2 and 3 (Fig. 6; Plate I; Perkins 2004, fig. 5). At the north-west corner, the foundation was traced about 12m southward from its junction with the north boundary wall (Fig. 6). The outside face of the main wall just south of this corner had subsequently been chosen to support a small new building (designated Building 2, see below) using the existing enclosure wall to form its eastern side. Perhaps significantly, a contemporary, internally projecting buttress was recorded some 9m from the north-west corner of the enclosure, in line with the south wall of this later building (Fig. 6). Another 12m-long section of this foundation was revealed adjacent to ROMAN VILLA, MINSTER-IN-THANET. PART VI: ENCLOSURE, BUILDINGS 2 & 5 343 the bath-suite, Building 3 (Perkins 2004, fig. 5, 3014). The foundation here showed evidence of a cream shelly mortar layer on its surface which must represent the bedding for the lost wall above. There were two more associated buttresses – an external one towards the south and an internal one 5m further north. Immediately to the north of the northern buttress, the foundation had subsequently been cut through by a late stone-lined drainage channel connected with Phase 3 of the adjacent baths (Perkins 2004, 44, fig. 5). An external boundary ditch ran westward from a point close the southern buttress (F. 3109/4018, see below). The North Boundary Wall The line of the north boundary wall was established in three area excavations (Figs 1, 2, 3 and 6), supplemented by intermediate trenches. More than half the total length of the foundation was exposed. It ran parallel with the north wall of the main villa house (Building 1), placed about 19m out from the north corridor (Fig. 1). At the north-west corner it joined with the contemporary foundation of the west boundary wall (Fig. 6, see above). At the north-east corner it probably originally butted onto the pre-existing Building 5, but the actual junction had not survived (Fig. General view of Building 2, looking north-east PLATE I KEITH PARFITT, EMMA BOAST ANDand gerald moody 344 Fig. 6 Detailed plan of Building 2 and adjacent features. Buttress ROMAN VILLA, MINSTER-IN-THANET. PART VI: ENCLOSURE, BUILDINGS 2 & 5 345 2, see above). For much of its length the main chalk footing was surfaced with small black pebbles. Two associated buttresses were identified – an external one in the central area and fragmentary traces of an internal one not far from the north-east corner (Figs 2 and 3). Abutting the north side of this foundation in the central sector was the remains of a later, more substantial mortared wall (9048) which probably related to a small building butted onto the outside of the boundary wall in this area. Sometime later, after this wall and the earlier villa enclosure wall had both been demolished, the boundary line was renewed with a substantial wooden fence (detailed below). Wall 9048 Abutting the north side of the chalk boundary wall foundation in the central sector was the remains of a later, more substantial e-w wall (Fig. 3, 9048). This had been significantly damaged by later pits, Fs 9038 and 9047, and its exact function is now unclear. Most probably, it formed the southern side of a small building, of unknown function, which extended from the outside face of the boundary wall. Where surviving, the wall was constructed of mortared flint and stood to a height of about 0.15m. It was at least 0.38m wide and was traced for a distance of 3.60m. It rested on a broader foundation composed of large chalk lumps set in brown clay. The foundation was traced for a distance of 4.20m. It was 0.88m wide and 0.30m deep (Fig. 7). It would seem that these remains represent the last traces of a small, though substantially built structure that once extended to the north of the boundary wall, perhaps as far as the edge of boundary ditch F. 9032/9045. A large pit had subsequently destroyed most of the structure (Figs 3 and 7, F. 9038, see below). This might conceivably represent a robber pit connected with the removal of the building. If this is correct, the depth of the pit suggests that the building must have included a basement of some kind. The surviving wall of the building had also been clipped by one of the post-holes (F. 9060) associated with the later fence-line (see below). The North Fence-Line The chalk foundation of the north villa enclosure wall had subsequently been cut along its northern edge by a series of post-holes (Figs 2 and 3, Fs 849, 833, 824, 831, 847, 848, 9050, 9052, 9054, 9056, 9058, 9060, 9062, 9064, 9066, 9068), spaced at intervals of between 1.00 and 3.75m. These post-holes were all circular or oval in shape, between 0.40 and 0.69m across and 0.17-0.38m deep. Most contained large flint packing stones and it seems clear they relate to a substantial timber fence-line that replaced most, if not all, of the old north villa enclosure wall after it had KEITH PARFITT, EMMA BOAST ANDand gerald moody 346 Fig. 7 Section across pit, F. 9038, wall 9048 and boundary ditches. ROMAN VILLA, MINSTER-IN-THANET. PART VI: ENCLOSURE, BUILDINGS 2 & 5 347 been demolished. Although late in the site sequence, the exact date of this fence-line is difficult determine due to an overall lack of associated datable material. Pit, F. 9038 This large pit was located immediately to the north of the north villa boundary wall (Fig. 3), cutting through the added wall 9048 and a substantial boundary ditch, F. 9032/9045 (see below). The pit was roughly pear-shaped and measured 6.85m (n-s) by 6.90m (e-w). Its sides were sloping and the base dipped gently down to the north (Fig. 7). The pit was between 0.60 and 1.00m deep and could just possibly represent a robber-pit associated with the removal of the structure represented by wall 9048 (see above). The silts filling this pit (Fig. 7, 9016 and 9024) contained very large amounts of painted plaster, together with about 180 pot-sherds. Much of the pottery dates to the mid/late second century but also includes a significant proportion of earlier, residual material. It would seem that this pit had served as a convenient dump for builder’s rubble, and some domestic rubbish, derived from the villa complex. Most of the painted plaster probably comes from an episode of extensive interior re-decoration, perhaps within the main house, Building 1. Some of the material, however, could represent debris derived from the building associated with the adjacent wall 9048, if this pit is related to its robbing (see above). Shaft, F. 821 Immediately north of Building 5 and the north wall of the villa enclosure, the top of a large shaft was located in 2002 (Fig. 2, F. 821). This was oval in shape and measured 4.70m (e-w) by 2.70m (n-s). It could only be safely excavated to a depth of 2.20m and the base was not reached. The impression gained was that there was still some considerable distance to go in order to reach the bottom of this feature and it seems quite possible that it represents the top of a well-shaft, similar to those located further down-slope, adjacent to Building 6A (Parfitt 2007, Fs 7105 and 7128; Fig. 1). The excavated filling of the shaft produced about 280 sherds of pottery, together with building debris, including tile, opus signinum and painted wall plaster, animal bone and marine shell. The lower levels exposed also appeared to include cess deposits. The pottery recovered ranges in date from the first to second centuries and includes some large, fresh pieces, as well as obviously residual material. Overall, the assemblage suggests that the shaft was backfilled between c.ad 120 and 140 (Lyne KEITH PARFITT, EMMA BOAST ANDand gerald moody 348 2006, Assemblage 10). After the shaft had been levelled, it was cut by one of the post-holes associated with the north fence-line (F. 824, see above), and also by a n-s ditch (F. 804, see below). Building 2 Sometime after the construction of the west boundary wall of the villa enclosure, a small structure had been butted onto its outside face some 6m south of the north-west corner (Figs 1 and 6). This has been designated Building 2. It was excavated in 1996, 1997 and again in 2001. The precise interpretation occasioned much debate on site and two alternative hypotheses, based on the presence of tile-lined channels, were considered – either a latrine or a corn-drying oven. That the structure actually represents a rectangular corn-drier of a type known from other Romano-British rural sites (Morris 1979, fig. 11) now seems certain. The channels located are the heating ducts which originally must have run below a small square structure that had been built against the existing villa enclosure wall (Fig. 6). Few other traces of this structure had survived subsequent activity. There was no evidence that an adjacent gate through the enclosure wall had existed to provide access from inside the villa enclosure. As surviving, Building 2 was defined by little more than the truncated remains of its tiled channels (Plate I). These must have originally been set under the floor of a structure around 4m square, with its eastern side formed by the pre-existing villa enclosure. The south-east corner had been completely removed by subsequent ploughing. The main furnace flue was positioned at the north-west corner, with the stoke-pit adjoining (Fig. 6, F. 1014). Extending from the furnace flue in the form of an inverted, reversed ‘F’ were two tile-lined channels, one continuing the line of the furnace flue southward before making a right-angled turn to the east, the other set at a right-angle to the east, extending as far as the enclosure wall. The channels were between 0.30 and 0.35m in width, their bases gently inclined at about 4 degrees to the horizontal, becoming shallower as they proceeded south and east (Plate I). They had been built in individual trenches cut into the underlying subsoil. The walls of the channels were constructed of tile including many broken fragments of tegulae, with some chalk blocks and flint nodules, all set in a coarse white mortar. The bases of the channels were formed by complete flat tiles, measuring 0.44 by 0.29m (corresponding with the Roman lydion brick). On the faces of the channels there were traces of opus signinum rendering. The deepest section of channel, near the furnace flue, was about 0.25m (4-5 courses). The sub-rectangular stoke-pit associated with Building 2 (Fig. 6, F. 1014) lay on the north side of the structure, cut into the terminal of an ROMAN VILLA, MINSTER-IN-THANET. PART VI: ENCLOSURE, BUILDINGS 2 & 5 349 earlier n-s ditch (F. 309, see below). The pit itself measured about 3.25m (n-s) by 1.60m (e-w) and was unlined. It was 0.50-0.70m deep with sloping sides and a flat base. At the southern end, adjoining the main flue, there was significant evidence of burning on the sides and base of the pit, with a thin layer of carbon lying over the base (Fig. 6). This confirms that there had been a fire positioned in this area and that the mortared channel which led away from it had been for the passage of hot air, rather than waste-water. After the stoke-pit had gone out of use it became filled by layers of silty clay loam containing some chalk, flint and tile fragments, together with a few pieces of painted plaster. Collectively, these fills produced 90 pot-sherds, including two pieces of samian ware, all of which fall within the period c.ad 120-250. The heating channels were filled with sandy soil which contained significant amounts of flint, tile and mortar rubble, much of which is likely to be derived from the structure above. There were also 128 pieces of pottery, datable to the second and third centuries, together with some animal bone. Overall, the pottery indicates that Building 2 had been abandoned and levelled by the end of the third century. The identification of Building 2 as the remains of a corn-drying oven, positioned outside the main villa enclosure seems beyond doubt. Although such structures are known from many Romano-British rural sites, their precise function has been the subject of considerable debate. It was originally assumed that they were used for parching grain prior to threshing and storage but recent experiments have suggested that such kilns were not very efficient in drying corn. It now seems more likely that many were used as malting kilns, where grain was spread out across a heated floor in order to encourage germination for the production of malt used in brewing beer. Corn-drying ovens are frequently found in the south and east of Britain and generally date to the late third or fourth century ad (Morris 1979, 20; Jones and Mattingly 1990, 230). For Building 2 at Minster, the available evidence suggests that it is earlier, contemporary with the main villa complex in use throughout the second and earlier third century. An example of a more typically late Roman corn-drier is provided by the structure recorded within Building 7 at Minster (to be described in detail in a future report, but see below). Boundary Ditches outside the Villa Enclosure A series of ditches was recorded outside the villa enclosure (Fig. 1). These are most likely to have delimited ditched fields and enclosures which lay beyond the main walled villa compound, although two (Fs 3163 and 7246) served as drains from bath-suites within Buildings 3 and 6A. At KEITH PARFITT, EMMA BOAST ANDand gerald moody 350 least some of the ditches located may have their origins in the pre-villa settlement known to have existed on the site (Parfitt et al. 2008, 311). F. 9032/9045, Ditch outside the north wall of villa enclosure Excavations in 2002 revealed part of a very substantial ditch running parallel with the north wall of the villa enclosure. This ditch lay some 4m to the north of the wall and was quite probably contemporary with it (Figs 1 and 3). It was traced for a minimum distance of 23.75m (e-w) and ranged in width from 3.10 to 3.75m. It was between 0.95 and 1.45m deep, with sloping sides and a dished base (Fig. 7). The filling produced 60 pot-sherds dating from the first to the early second century ad, suggesting that the ditch belongs early in the villa sequence. Sometime after the ditch had been in-filled it was cut into by two large pits (Figs 3 and 7, Fs 9038 and 9075) which produced some second-century pottery along with significant amounts of earlier, residual material. Pit, F. 9038, perhaps a robber pit, has been described above. Pit, F. 9075, followed the line of the ditch and may actually represent a re-cut of F. 9032/9045 (Fig. 3). It produced significant quantities of pottery (300 sherds), painted plaster, tile and animal bone. F. 9036, Later ditch outside the north wall of villa enclosure Once pits F. 9038 and F. 9075 (see above) had been backfilled, a substantial new ditch (F. 9036) was cut following the line of the earlier ditch, F. 9032/9045 (Fig. 3). This new ditch was 1.50-2.10m wide and 0.85-1.00m deep, with sloping sides and a rounded base (Fig. 7). Its lower filling produced 220 pot-sherds datable to the late first/late second century. The upper filling produced a further 300 sherds mostly dating to the second and third centuries but also including about 50 pieces of late third/fourth-century date. Significant quantities of animal bone and tile were also recovered and it would seem that this ditch was open throughout the main occupation of the villa complex, serving as a replacement to the previous ditch in this area. The late Roman material recovered suggests that it remained partially open after the main villa complex had been abandoned. A parallel shallow gully (F. 9015) located immediately to the south of this ditch may have been open at the same time (Figs 3 and 7). F. 804, North-south ditch cutting line of north wall of the villa enclosure Towards the north-east corner of the villa enclosure, the line of its north wall together with the top of a nearby shaft, F. 821 (see above) had subsequently been cut across by a straight ditch running n-s for an unknown distance (Figs 1 and 2, F. 804). This ditch passed through the ROMAN VILLA, MINSTER-IN-THANET. PART VI: ENCLOSURE, BUILDINGS 2 & 5 351 gap between the north enclosure wall and Building 5 so that no definite relationships could be established, but it clearly post-dated the shaft, in-filled during the mid-second century (see above). The purpose of this ditch remains uncertain. The ditch was 0.60-1.20m wide and 0.40m deep, with sloping sides and a flat base. It was traced for a total distance of about 16.50m, of which 11.50m lay inside the villa enclosure. Deposits of silty loam filled the ditch and produced almost 140 pot-sherds datable to the period c.ad 150-250. Taken with the date for the in-filling of the shaft, the ditch was probably cut in the later second century and back-filled by the middle of the third century. F. 309, Ditch outside NW corner of Villa enclosure This ditch ran northwards from a point just outside the north-west corner of the villa enclosure wall (Figs 1 and 6). It was traced for a minimum distance of about 4m from its southern terminal but its full extent was not revealed. A slight easterly curve in its line could suggest that it made a right-angled turn beyond the excavation limit to run parallel with the north villa boundary wall. The ditch was 0.75-1.00m wide and less than a metre deep, with sloping sides and a dished base. The southern terminal had subsequently been dug into and expanded to create the stoke-pit (F.1014) for a corn-drier (Building 2, see above). The rest of the ditch seems to have been back-filled by this stage. F. 3109/4018 and F. 3108/4014, Re-cut ditch, west side of villa enclosure This ditch extended westwards from the west wall of the villa enclosure, opposite the bath building (Building 3). It was traced in the 1998 and 1999 seasons for a total distance of some 15m within the excavation and clearly continued beyond, heading towards the adjacent stream valley (Fig. 1; Perkins 2004, fig. 5 for detailed plan). As originally cut, the ditch (F. 3109/4018) was up to 1.95m wide and 0.85m deep, with steeply sloping sides and a rounded base. It was filled with a series of deposits which collectively produced a quantity of domestic rubbish, including more than 400 pot-sherds, mostly datable to the period c.ad 120-200. The lower filling also contained the complete skeleton of a young horse (Plate II, 4117). Sometime after the original ditch had become full it was re-cut on the same axis, slightly further to the south (F. 3108/4014; see Perkins 2004, fig. 5). The new ditch was between 0.70 and 1.60m in width and up to 0.60m deep, with steeply sloping sides and a rounded base. It was filled with deposits of soil and rubble which together produced about 180 pot- KEITH PARFITT, EMMA BOAST ANDand gerald moody 352 PLATE II sherds, mostly of similar dates to the material from the previous ditch. Many pieces are likely to have weathered out of the filling of the original ditch but there is nothing to suggest that this re-cut ditch remained open much after the mid-third century. Although situated close to the bath-house (Building 3), there is no evidence that this ditch was ever connected with the drainage of waste-water from that structure (Fig. 1). A separate stone-lined channel and another extra-mural open ditch heading north seem to have served this Horse burial in ditch, F. 3109/4018 ROMAN VILLA, MINSTER-IN-THANET. PART VI: ENCLOSURE, BUILDINGS 2 & 5 353 function (Perkins 2004, fig. 5, F. 3163), making it more likely that the present ditches related to a field boundary that ran up to the villa enclosure wall. The pottery dating evidence suggests that the ditch was finally in-filled sometime during the third century ad. F. 5003/5014, Ditch outside NE corner of Building 4 Extending eastwards from the presumed site of the destroyed south-east corner of the villa enclosure wall (see above) was a substantial ditch (Fig. 1). In its earliest form (F. 5003) this was probably contemporary with the walled villa enclosure and seems to have continued the line of its south wall. The development of the corridor house here, Building 4, eventually required the original ditch terminal to be back-filled to make-way for its new eastern corridor wall. The line of the ditch, however, was maintained and re-cut (as F. 5014), now terminating just outside the new corridor wall. The pottery dating evidence suggests that the ditch was finally back-filled around the middle of the third century ad (Parfitt 2006, 129, figs 2 and 3). F. 7246/7233, Ditch running SW from Building 6 At the south-west corner of the villa complex an open ditch (F. 7246) originally took waste-water away from the plunge pool within the bath suite of Building 6A (Room 41). This ditch ran downhill to the south-west and was traced for a distance of some 12m in the excavation (Fig. 1). After the baths ceased to function and a drain was no longer required here, the ditch was re-cut and retained as a shallow boundary ditch following the same line (F. 7233). The pottery dating evidence again suggests that the ditch was finally in-filled sometime during the third century (Parfitt 2007, 288, figs 2 and 6). discussion of the villa enclosure Although many Romano-British villa sites had enclosing walls, fences or ditches, the compound at Minster is particularly neat and regular. The main house (Building 1) and its adjacent bath-suite (Building 3) were set entirely within the walled area, positioned at the top of the slope overlooking all the lower parts of the enclosed compound to the south (Fig. 1). Approached from the Wantsum shore to the south, the villa house must have appeared very prominently on the skyline at the top of the hill. Buildings 2, 4 and 6 were butted onto the outside of the enclosure and it is no doubt significant that the two largest structures at the lower end of the enclosure, Buildings 4 and 6, stood immediately outside the projected lines of the side walls of the main house, so that from a distance the KEITH PARFITT, EMMA BOAST ANDand gerald moody 354 buildings would have presented a continuous elevation, with the principal structure standing higher in the central sector. In detail, the layout of the enclosure at Minster (Fig. 1) is a rather less common arrangement than sites where lengths of boundary wall or fence ran between, and connected to, the principal buildings (e.g. Darenth, Keston, Brading and Sparsholt). The enclosure around the winged-corridor villa at Ditchley in Oxfordshire, provides a closer parallel to the Minster arrangement (Ralegh Radford 1936, fig. 8). The Ditchley boundary wall delimited an enclosure rather more square in plan than that of Abbey Farm but covering about the same area. At Ditchley, an outer ditch surrounded the enclosure wall. At Minster, a ditch in a comparable position, renewed several times, occurred along the north side but this did not continue around the other sides of the villa enclosure. The precise date of the construction of the villa enclosure at Minster remains uncertain. There is little helpful dating evidence, although the walls clearly pre-dated Buildings 4 and 6A, as well as Building 2 (see above) and the late stone-lined drain leading from Building 3. On balance, it perhaps seems most likely that the boundary walls were first laid out at the same time as the main villa house and are quite likely to be contemporary with Phase 1 of Building 1, dated to the late first century (Parfitt et al. 2008, 331). Confirmation that this was the case is possibly demonstrated by the similarity of the rammed chalk and pebble construction of both sets of foundations. If the villa enclosure walls are this early, it must indicate that the preceding pebble foundations (Building 5, etc.) located at the north end represent pre-villa structures. From the surviving foundations it is difficult to gauge both the original height and the appearance of the villa enclosure wall itself. A figure of somewhere between 1.50 and 3.00m for its height is perhaps most likely but the nature of its construction must remain largely unknown. That it was stone-built seems to be implied by the presence of mortar on top of the footings along the west side (see above). The main wall need not have been plain and might well have been plastered or painted; it could even perhaps have incorporated ornamental features in its construction. To preserve sensible proportions, the east and west side-walls of the enclosure must have been stepped to take account of the sloping hillside upon which the site was set (Fig. 1). It seems possible that some of the buttresses recorded along these side-walls mark the positions of changes in wall height. The narrow, relatively insubstantial nature of its foundations, indicate that the villa enclosure wall could not have been of sufficient proportions to have had any sort of defensive function, beyond keeping out the occasional vagrant or wild animal. This was also considered to be the case with the enclosure at Ditchley (Ralegh Radford 1936, 41). Both enclosures thus stand in marked contrast to the massive defensive wall ROMAN VILLA, MINSTER-IN-THANET. PART VI: ENCLOSURE, BUILDINGS 2 & 5 355 that surrounded the late villa complex at Gatcombe in Somerset (Branigan 1977). Most probably, the enclosure at Minster served to define either a formal courtyard/garden in front of the main house, a more work-a-day farmyard, or some combination of the two. Nothing specific was recorded at Abbey Farm to help establish the nature of this courtyard but the care with which the various buildings on the site had originally been laid out, following a closely symmetrical design, tends to imply that the area in front of the main house, in its heyday, was not a rough yard occupied by scratching hens and goats but a formally laid out garden designed as an extension of the overall architectural design of the complex. That the area enclosed by the boundary wall at Minster was not a perfect rectangle is readily apparent from the site plan (Fig. 1). In a number of Romano-British villa complexes enclosing walls and out-buildings were specifically designed to converge towards the main house in order to give additional emphasis to this principal structure. It has been suggested that such fan-shaped layouts are an architectural feature imported from villas in northern Gaul (Black 1987, 140; Johnston 2004, 24). At Minster, however, the enclosed area seems to have been only a few metres wider at the southern end of the compound in comparison with the north. This would hardly have been noticeable to any visitor approaching the main house. The natural fall of the ground within the villa enclosure, particularly at the south-western corner, is quite marked (Fig. 1) and it seems most likely that it was the existence of this slope that led to inaccurate surveying during the initial laying out of the main villa house and its enclosure. Interestingly, there is a similar amount of distortion between the house and the enclosure walls, implying that the boundary wall lines were established by measuring out set distances from the walls of Building 1, which itself had been laid out with a marked skew (Parfitt et al. 2008, 313). Although not examined extensively, a sufficient number of ditches extending out from the villa boundary walls were recorded to suggest that the walled enclosure around the villa complex at Minster was surrounded by ditched fields and paddocks. On the evidence of other, more extensively excavated sites, such boundary ditches are likely to have covered a considerable area, although no obvious traces of them have been recorded on any local aerial photographs of Minster. These ditches need not all be exactly contemporary and some certainly replaced earlier ones connected with a pre-villa settlement (Parfitt et al. 2008, 311). The ditches investigated do not seem to have been maintained after the main occupation of the villa complex had ceased, around the middle of the third century, suggesting that there were some major changes in local land-division after this time. KEITH PARFITT, EMMA BOAST ANDand gerald moody 356 The combined site evidence indicates that the entire villa complex at Minster had been given up sometime during the third century, with all the principal buildings subsequently robbed for their re-usable tile and building stone. This may also have been the case with the villa enclosure wall. Nevertheless, there is evidence for renewed late Roman activity on the site (Holman and Parfitt 2005, 210) when a major new timber building containing a large corn-drier was erected in the middle of the former villa enclosure (Building 7, to be detailed in a future report). This re-use of the abandoned site raises the possibility that the old villa enclosure walls might still have survived, even if in an over-grown and partially decayed state, and now served to delimit the working area around the new structure. Recorded evidence for late changes to the villa boundary walls include the re-modelling of the south gate sometime after the start of the third century and the replacement of the north wall with a stout wooden fence. Such alterations could be associated with this later re-use of the site, although there is insufficient dating evidence to confirm this. acknowledgements The Abbey Farm Villa excavation team again wish to express their gratitude to the Kent Archaeological Society for the funding and administration of the excavations. In particular, the then President, Paul Oldham, must be warmly thanked, together with David Bacchus and Chris Pout of the Fieldwork Committee. Farmer Mr Jack Clifton arranged access to the site and keenly followed developments throughout the course of the work. Thanks are also due to Minster Parish Council, and the staff of Minster Agricultural Museum for their kind co-operation. Members of Thanet Archaeological Society have spent much time processing and cataloguing the finds. The Project Directors, Dave Perkins and Keith Parfitt, would like to thank the Supervising Archaeologists, Emma Boast, Tim Allen, and Ges Moody, for their support and enthusiasm during all phases of excavation and post-excavation work. Susan Deacon, Emma Boast, Oliver Gardner and Ges Moody prepared the site drawings for publication. Finally, the hard work of all the excavators must be recorded – little would have been achieved without their efforts. bibliography Black, E.W., 1987, The Roman Villas of South-East England (B.A.R. British Series 171). Branigan, K., 1977, Gatcombe Roman Villa (B.A.R. British Series 44). Holman, D.J. and Parfitt, K., 2005, ‘The Roman villa at Minster-in-Thanet. Part 2: The late Iron-Age, Roman and Later Coinage’, Archaeologia Cantiana, cxxv, 203-228. ROMAN VILLA, MINSTER-IN-THANET. PART VI: ENCLOSURE, BUILDINGS 2 & 5 357 Johnston, D.E., 2004, Roman Villas (Shire, 5th edition). Jones, B. and Mattingly, D., 1990, An Atlas of Roman Britain (Oxford). Lyne, M., 2006, ‘The Pottery from the Minster Roman Villa’ (archive report commissioned by the KAS). Morris, P., 1979, Agricultural buildings in Roman Britain (B.A.R. British Series 70). Parfitt, K., 2006, ‘The Roman Villa at Minster-in-Thanet. Part 3: The Corridor House, Building 4’, Archaeologia Cantiana, cxxvi, 115-133. Parfitt, K., 2007, ‘The Roman Villa at Minster-in-Thanet. Part 4: The South-West Buildings, 6A and 6B’, Archaeologia Cantiana, cxxvii, 261-296. Parfitt, K., Perkins, D., Boast, E. and Moody, G., 2008, ‘The Roman Villa at Minster-in-Thanet. Part 5: The Main House, Building 1’, Archaeologia Cantiana, cxxviii, 309-334. Perkins, D.R.J., 2004, ‘The Roman Villa at Minster-in-Thanet. Part 1: Introduction and Report on the Bath-House’, Archaeologia Cantiana, cxxiv, 25-49. Ralegh Radford, C.A., 1936, ‘The Roman Villa at Ditchley, Oxon’, Oxoniensia, 1, 1-42.