The Roman Villa at Minster-in-Thanet. Part 1: Introduction and Report on the Bath-House

THE ROMAN VILLA AT MINSTER-IN-THANET. PART 1: INTRODUCTION AND REPORT ON THE BATH-HOUSE D.R.J. PERKINS In north-east Kent, on the Isle of Thanet, a knowledgeable observer scanning the structure of Minster's oldest buildings, the Abbey and church of St Mary's, would deduce the presence nearby of Roman building remains, as the re-use of Roman materials, mainly tile, is very evident (Jones 1992, 234 ). Two hundred years ago the ruins were there to see. W.H. Ireland (1828, p. 481) recorded that: 'In a field called "Twenty Acres", about a quarter of a mile eastward of Minster, a variety of foundations are apparent, as if some chapel or oratory had originally existed there'. By the last half of the twentieth century these remains had been so denuded that only a stony patch could be seen. Farmer Jack Clifton was told by older farm hands, and accepted, that this marked the site of a demolished pig-pen. However, under ideal conditions, the 'pig-pen' exhibited an extensive and explicit negative crop-mark of a winged Roman villa. In 1979 this was photographed and plotted by HBMC (Neg. no. NMR 1661, Frame 363 ), from whence, via the office of the County Archaeologist, the data came to the attention of Thanet Archaeological Society (Perkins 1996). The site lies on Abbey Farm, NGR TR 3135 6463 (Figs 1-3), and was made the subject of a field-walking survey at the first opportunity. It was also prospected by a local metal detectorist group, who found a tight cluster of about 40 Roman coins. In 1991 Jack Clifton kindly gave permission for an evaluation of the site by hand-cut trenching and this was carried out by Society members. Results from four trenches indicated that substantial building remains survived, but that they were very close to the surface and subject to regular plough attrition. A problem was therefore presented in that to record the remains ahead of their destruction would require a lengthy and extensive excavation; one for which no source of funding came to mind. The difficulty was resolved in 1996, when the site was adopted by the 25 Isle of Thanet Ramsgate bbsfleet Villa (J Richborough Fig. l Map of north-east Kent showing location of Abbey Farm Roman villa in relation to the Wants um Channel, Roman roads on Thanet and other major Roman sites in the area. N BA - Bronze Age R·B - Romano British AS - Anglo Saxon Jl·B Settlement & uria\s .a. R-B Settlement & Bunals 􀂒B lr rk - '\. i I - onwo ngs R-B Buri􀂓 50rn _ .. R-B Cemetery/AS Burials -----40m........- __ 􀀋 ACoinHoard J 1 20􀂔 BA Burial Abbey Fann Villa ( ( _ J 􀀅 ...,, J"\..& R-B Settlement 􀂕,AM . G 􀂖-􀂗--ds St.M􀂘Paris h Chu.rch,A ones􀂙􀂚::? i s po 􀂛 I Oni Wantsum Ch a.t1iiet · R-B/A S Buri.als Minster Parish Boundary ,A Ebbsfleet Villa Med. -Medieval ,A R-8 Cemetery 0 􀀄 3km Fig. 2 Map showing the location of Abbey Farm Roman villa in relation to Minster parish boundary, and other local archaeological sites. D.R.J. PERK.INS Fieldwork Committee of the Kent Archaeological Society as the venue for a series of annual training excavations (Perkins 1996). These were well attended, both by enrolled trainees, and experienced amateurs, mainly from Thanet Archaeological Society (patrons of the project) and members of the Dover Archaeological Group. Training excavations (Plate I), each of fourteen days duration, took place in the years from 1996 to 1999 (Perkins 1996; 1999), and were recommenced in 2001. Research is ongoing. To date, excavation and resistivity survey has revealed part of a large villa complex. To examine the complex in its entirety, assuming that investigations continue, could take at least several more years, and generate a publication text well beyond the capacity of a single volume of Archaeologia Cantiana. The object of this paper, therefore, is to commence publication of the villa complex and its finds by instalment, initially dealing with the site's location and describing the detached bath-house, Building 3, situated immediately to the west of the main villa house. On the southern side of the Isle of Thanet, just inland from what was once the shore of the Wantsum Channel there are deep measures ofThanet Beds sands, forming foothills to the higher chalk downland plateau that occupies the central part of the island. One such sandy hilltop, overlooking the modern village of Minster, was chosen for the site of the Roman villa complex (Figs 1 and 2). The hilltop itself is almost flat and stands at an elevation of about 16m OD. Immediately to the west lies a small, narrow valley (Fig. 3). In its bottom an emerging spring feeds a little stream which supplies a line of fish ponds associated with the nearby medieval grange (now Minster Abbey), originally belonging to St Augustine's Abbey. This stream must have provided an important source of fresh running water (rare across much of Thanet) from prehistoric to medieval times. A ridge of chalk downland, forming part of Than et' s central plateau, rises above the villa site and this ridge is followed by a prehist· oric trackway, which eventually became known as 'Dunstrete' (Fig. 1). Ceramic and lithic evidence obtained during excavation and survey of the villa site have established that the hilltop also attracted some settlement during the Neolithic, late Bronze Age, and late Iron Age periods. The Roman villa was thus established within a landscape that had long been occupied (Fig. 2). If a splendid view was valued by the villa builders (as is likely), they could not have been better served. West and south-west were the broad waters of the Wantsum sea channel (Fig. 1 ), with the town and fortress of Rutupiae (Richborough) upon an island some 4.5 km to the south. Beyond, as a backdrop, was the Kentish mainland, constituting the heights of the North Downs from Walmer to Dover. More practical considerations were undoubtedly the proximity of the 28 MINSTER-IN-THANET ROMAN VILLA : INTRODUCTION AND BATH-HOUSE Fig. 3 Abbey Fann Roman villa in relation to local contours, the stream and medieval fish ponds. 29 D.R.J. PERKINS spring-fed stream and the Wantsum, with a safe natural harbour at the foot of the hill about 500m away (Fig. 2). Other Roman building remains of various kinds recorded elsewhere on the Isle of Thanet have established that the Minster villa complex was far from isolated. At the time of writing there are eighteen sites on Thanet that can be tentatively or positively identified as Roman villas. Details of these have been previously summarised (Perkins 2001, appendix 1 ). In spite of what might seem an impressive list, the dearth of real evidence about Roman Thanet is best illustrated by the fact only two of the listed villa sites - the present site and Ebbsfleet (Fig. 2) have been sufficiently investigated to provide stratified dating evidence. It is, however, now possible at least to refute the long held contention that Roman Thanet was a sparsely populated 'granary' for Rutupiae and Regulbium (Reculver), the two fortresses on the west bank of the Wantsum (Fig. I). Within their protective shadow, the Isle actually seems to have held a flourishing community. To date, the excavations have succeeded in locating and substantially excavating six separate buildings (Fig. 4, Buildings 1-6) and it seems likely that others still remain to be discovered, together with a variety of ancillary structures. The heavily plough-damaged remains of the main villa house is represented by Building I, the structure initially revealed on the air photograph (see above). This south-facing building is some 40m in length and comprises a main E-W range of seven rooms. From the ends of this, wings, each of two rooms, extend south, the whole being enclosed by flanking corridors (Perkins 1996; Fig. 4). A number of phases of additions and modifications were apparent within the surviving structure, implying a quite complex building history, which is still being analysed. Immediately to the west of Building I lay a small detached bathhouse (Building 3; Figs 4 and 5). This was fully excavated in 1998 and is described in detail below. Both Buildings 1 and 3 were placed at the northern (higher) end of a large, walled enclosure (Figs 3 & 4). In plan, this enclosure was not quite a true rectangle (Fig. 4 ), measuring 67- 73m (E-W) by l l 7- I 22m (N-S). Evidence for a gateway, marked by large timber posts, was located in the middle of its south wall, although this may not have been an original feature. Rough metalling extended north and south of this entrance and must represent an access road. A rare silver coin of Septimus Severus was recovered from this metalling suggesting that it was not }aid before the start of the third century. Set on the outside of the west wall of the enclosure, at the north corner, lay Building 2 (Fig. 4). This small, rectangular structure 30 MINSTER-IN-THANET ROMAN VILLA: INTRODUCTION AND BATH-HOUSE f B2 0 50m 0 150ft Fig. 4 Plan of the villa complex showing overall layout and the position of excavated buildings (B3, bath-house). 31 PLATE I Excavation work in progress (photo D. Perkins) MINSTER-IN-THANET ROMAN VILLA : INTRODUCTION AND BATH-HOUSE contained tile-lined channels and has been interpreted as either an elaborate corn-drying oven/malting kiln, or a latrine. Another small structure of uncertain purpose, located at the north-east corner of the walled enclosure but apparently predating it, is represented by Building 5. Building 4 consists of a stone-built house, situated some 7 Sm downhill (south) of the main villa. It lay immediately outside the enclosure wall, at the south-east corner and several phases of development were again apparent. A corresponding structure outside the south-west corner (Building 6) is presently under excavation. The excavations and survey work have produced large quantities of pottery, tile, glass, animal bone, finely painted plaster, mosaic fragments, coins and other finds. These will all be detailed in future reports, once the excavations have been completed. Dating of the excavated pottery and coins has been undertaken to allow an outline chronology for the site to be established and the results have been incorporated into the present report. From the available evidence, it would seem that the villa complex was established towards the end of the first century AD, with the main period of occupation during the second century and continuing into the early third century. The site appears to have been abandoned by the late third century, however, raising the possibility of the existence of a separate, late Roman villa complex not far away. THE BATH-HOUSE EXCAVATION by Keith Pa,fitt A detached bath-house (Building 3) was located within the walled villa enclosure, immediately to the west of the main house (Building I). Although there was less than a 2m gap between the buildings, they represent separate structures (Figs 4, 5 & 6). No clear stratigraphic relationship could be established but the bath-house had been placed in line with the end of an added corridor around the west wing of Building I (Fig. 5), implying that the main house was already in existence. Because of their close proximity, some sort of roofed connecting passage could easily have been constructed between the two buildings so that the baths would have effectively formed part of the same overall complex. The bath-house comprised a comparatively small, rectangular building, measuring 9.55m (N-S) by 7.15m (E-W), with an external apse on the west side (Fig. 5 and Plate II). It was first discovered during the 1997 season and was fully excavated the following year. The pottery and coin evidence suggest that the structure was first erected sometime during the late first or early second century and it was probably abandoned in the early third century. Subsequently, it was extensively 33 w .i,.. Recutditch F.3108 F. 3109 - Phase l E2:1 Phase2 􀀁 Phase3 0 10 20 30ft I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 0 I S.7 Fig. 5 Detailed plan of the bath-house (Building 3). Building I :-- "Cl ('TJ F. 1020/ 3084 MINSTER-IN-THANET ROMAN VILLA: INTRODUCTION AND BATH-HOUSE robbed and later suffered from plough damage, so that no upper floors or occupation levels survived and all the walls had been removed to foundation level. No evidence for the position of any doorways survived. The demolition deposits infilling the deeper, hypocausted rooms produced quantities of painted plaster and tesserae, demonstrating that parts of the bath-house originally had brightly painted walls and plain red tessellated floors. Around thirty pieces of glass must have come from windows in the building. Although the preservation of the building was relatively poor, analysis of the surviving fabric indicated that there had been several phases of development (Fig. 5), with the original building (Phase 1) subsequently undergoing some major refurbishments and modifications (Phases 2A, 2B & 3). Phase 2A saw the rebuilding of the hypocaust system and the enlargement of the main Hot Room. At about the same time a new external, stone-lined drain connected to a latrine pit was added along the north and part of the east side of the building (Phase 2B). In Phase 3 this drain channel was extended westwards through the villa boundary wall. Phase 1, The Original Bath-house (Fig. 5 and Plate II) In its earliest form the bath-house contained seven rooms arranged in two rows on either side of a longitudinal dividing wall (Rooms 20-26), with a small projecting apsidal room on the west side (Room 27). Rooms 21, 24, 25 and 26, on the south and west sides of the main block, were all basemented and associated with the remains of a hypocaust system. Room 24 was the furnace room. Rooms 22 and 23, on the east side, had never been hypocausted but no original floor levels survived. Although sunken, Room 20 did not seem to be connected to the hypocausted rooms and was probably a cold plunge-bath (see below). All the walls associated with the main Phase 1 construction were well built of large flint cobbles with occasional blocks of tufa and tile levelling courses (?discontinuous), set in a white-pink gritty mortar. Sub-floors of opus signinum [3153 + 3033] survived in the base of hypo- causted Rooms 21, 25 and 26 (Fig. 7, Section No. 7) and these are also likely to have been laid in Phase I. Description and Identification of the Rooms Room 20 This was situated at the north-east corner of the building (Fig. 5). It was a deep room but there was no evidence that it had ever been hypocausted (Fig. 6, Section No. 5). More probably, it represents a sunken cold plunge-bath with a contemporary tile-edged 35 S.5 e S.6 s 1997 Bockll 200 3044 3046 30<7 3050 3051 3161 I 3158 3162 3125 0 Finl b&e3127 Fig. 6 Sections across the bath-house (Building 3). Section Nos. 5 & 6. w 2m S.4 F. 1020 S.7 s W N r-- 3017 :l019 S.9 3044 3079 3073 N 0 Fig. 7 Sections across the bath-house (Building 3). Section Nos. 4, 7 & 9. 􀀃 lz s Ul -l tTl :;o 2 .:..i ::i:: 3130 􀀄 􀀅 N :,;, 0 􀀃 ► z < p 􀀆 􀀇 :;o 0 0 C 􀀈 0 z )> z 0 to 2m )> -l ::i:: ::c 0 C Ul tTl PLATE II Photograph of the excavated bath-house (photo D. Perkins) MINSTER-IN-THANET ROMAN VILLA: INTRODUCTION AND BATH-HOUSE aperture in the north wall forming its drain. It is likely that this drain originally discharged into a large open ditch running along the northern side of the building and through the villa enclosure wall to the west. During Phase 2B this ditch had been partially replaced by a stone-lined drainage channel, constructed within it (see below). The floor of Room 20 had been robbed but a series of early deposits seemed to represent undisturbed, pre-floor levelling layers. Resting on the natural clay at the base of the construction pit was a thin mortar layer (Fig. 6, Section No. 5, context 3050), apparently associated with the original building of the walls. It produced eight sherds from a lid-seated bead rim storage jar dating to the second half of the first century AD (Assemblage A, see below). A clay dump layer above [3049], containing a few pieces of tesserae, was sealed by another thin construction layer of crushed opus signinum [3 I 30]. This was followed by a further -clay levelling layer [3048] which may have supported the final floor (Fig. 6, Section No. 5). It produced a single pot-sherd in the form of a lid datable to the late first-second century AD, together with a few pieces of tesserae and painted wall plaster. Uncertainty remains as to whether the clay dump deposits relate to the first phase of the room's use, or whether they are associated with later alterations and changes in floor level. Room 21 This was a square hypocausted room situated at the southeast corner of the building (Figs 5 & 7). Since it lay at the greatest distance from the furnace, it seems most likely that it originally constituted the Warm Room (tepidarium) of the baths. During Phase 2A the dividing wall between Rooms 21 and 26 was demolished to foundation level and a continuous new hypocaust system built across its line. Room 21 then seems to have become a continuation of the Hot Room (Room 26, see below). Sometime after the abandonment of the bath-building the outer walls and foundations of Room 21 were completely robbed away [F. 1020/3084]. Room 22 This was an unhypocausted room, almost centrally placed on the eastern side of the building (Figs 5 & 7). From its positioning in relation to the heated rooms it could be interpreted as a rather small Cold Room (frigidarium), but its size would be more appropriate for an entrance lobby facing the main house. No original floors had survived but clay levelling layers between the walls (Fig. 7, Section No. 4, contexts 3027 & 3029), produced fragments of tile and painted plaster and a sherd from a necked jar dated c. AD 70-170. Room 23 This was another unheated room on the east side of the building (Figs 5 & 7). It may have served as the Undressing Room (apodyter- 39 D.R.J. PERKJNS ium) but its size in comparison to Room 22 and position next to the probable cold plunge-bath (Room 20), could suggest that it constituted the Cold Room (frigidarium) of the baths. Again, no original floor levels had survived but a clay levelling layer abutting the walls (Fig. 7, Section No. 4, context 3032) produced fragments of tile and painted plaster. At some later stage a square, stone-lined pit [3151], about 0.50m across, had been inserted into the north-eastern corner of the room (Fig. 5). This was connected to a long external stone-lined channel running around the northern part of the building (see below, Phase 2B). Fairly certainly, the pit represents a latrine. The associated stone channel implies that the pit was periodically flushed with running water, presumably somehow diverted from the outlet of the cold plunge-bath in Room 20. Rooms 22 and 23 both appear to be rather small for their suggested functions. It is possible that they were actually linked to form a single, larger room, with the dividing foundation serving merely to take the weight of a broad connecting arch. Room 24 Situated at the north-west corner, this was the largest room within the bath-house and clearly formed the furnace room (praefurnium) for the hypocaust system, with the furnace itself set against the south wall (Figs 5-7). During Phase I a central tile-edged aperture in the south wall, 1.08m wide, had conducted the heat from the furnace into the baths. Subsequently, the entire structure was remodelled and the central opening was narrowed to about 0.60m (Phase 2A). Most of the other structural features recorded within this room also seemed to relate to the working arrangements of the later baths (Phase 2A, see below). Room 25 This lay immediately south of the furnace chamber (Room 24) and represented the smallest room within the building (Figs 5 & 7). The east and west walls were markedly thicker than the walls of all the other rooms. A central flue, lined with broken tiles set in white mortar [3157] and about 0.60m wide, linked the furnace in Room 24 with Room 26 (Hot Room) but this appeared to be a late feature. The original layout was unclear. From its small size and location adjacent to the furnace, this room may be suggested as a hot plunge-bath. Gully [F. 3013) could represent the associated external drain from this bath, rather than Room 27 (see below). Room 26 This square room was situated in the south-west corner of the bath building (Figs 5 & 7) and from its position in relation to the 40 MINSTER-IN-THANET ROMAN VILLA: INTRODUCTION AND BATH-HOUSE furnace there can be little doubt that it represents the main Hot Room (caldarium). Sometime after its original construction, the east wall had been completely removed and the hypocaust structure substantially remodelled to create a much larger Hot Room occupying all the south end of the building (Phase 2A, Room 21/26, see below). Room 27 This was a small apsidal room situated on the western side of Room 26 (Fig. 5). It appeared to form part of the original structure but its function is not certain. From the shape and position it is possible that it represents a hot plunge-bath but the relatively small size would have meant that it could not have been used by more than one or two individuals at a time; nor was there any evidence that the room had been connected to the hypocaust system. An alternative identification, as a schola, typically found in many baths, can be suggested as being equally likely. This would have contained a water basin (Labrum) for use by bathers. A shallow external gully [F. 30 I 3] leading northwards, away from the apse, may represent an associated drain (Fig. 5), such as would have been required if the room did contain a plunge-bath. A schola, however, is unlikely to have needed such a drain and if this was the function of Room 27, it seems more probable that the gully actually took waste-water away from a plunge-bath which occupied Room 25. The gully appears to have discharged into a deeper ditch that served to take waste-water from the cold plunge-bath in Room 20 (see above). During Phase 3 gully [F. 3013] was blocked by an extension of the stone lined drain which had replaced the main ditch (see below). TABLE I. DETAILS OF ROOMS WITHIN THE BATH-HOUSE Room 20 21 22 23 24 25 26 27 21/26 Internal size (m) Sunken E-W x N-S 2.75 X 1.25 Y C. 2.70 X C. 2.70 Y 2.70 x 1.23 N 2.75 x 1.85 N 2.70 X 3.55 Y 2.12 X 1.15 Y 2.75 X 2.70 Y 0.62 X 1.55 N C. 6.00 X 2.70 Y 41 Identification Cold plunge-bath Warm room (tepidarium) ?Cold room (frigidarium) ?Undressing room & latrine Furnace room (praefurnium) Hot plunge-bath Hot room (caldarium), Phase I Hot plunge-bath, or schola enlarged Hot room, Phase 2A D.R.J. PERKINS Phase 2 Changes to Heating and Drainage At some stage, the bath-house underwent major alterations and modifications. The hypocaust system was rebuilt in Phase 2A, leaving the details of the original arrangement less than clear in many places. The drainage from the building was improved when a new latrine was added in Room 23 during Phase 2B. Whether all these changes occurred at the same time remains uncertain but it seems quite likely. Phase 2A, Later Hypocaust Arrangements The hypocaust system of the bath-house had been completely reconstructed, apparently to accommodate a major extension of the Hot Room (Fig. 5). During Phase 2A the length of the Hot Room was more than doubled, to about 6m, by the removal of the original Phase 1 dividing wall between Rooms 21 and 26. In order to conduct hot air direct from the furnace chamber to the extended Hot Room (Room 21/26) a long central, tile-lined flue was construction below Room 25. The extension of the Hot Room could indicate a larger number of people were now using the baths, or perhaps it reflects the rising status of the villa occupants. The suspended hypocaust floor of Room 21 /26 had subsequently been robbed (Fig. 7, Section Nos. 4 & 7) but it was clear that it had originally been supported on flat-topped, sloping-sided 'ramps' set around the side-walls, and central pilae bonded in clay (Section No. 7). Both the ramps and the pilae stacks extended across the demolished remains of the wall that had originally divided Rooms 21 and 26, so all must relate to the later heating system. The central pilae stacks were arranged in the three rows (Fig. 5). The middle row rested directly on the opus signinum basement floor, whilst the outer rows were set into the sloping sides of the wall ramps. The wall ramps, themselves, were constructed from flint cobbles and re-used building rubble, set in cream-brown clay with crushed mortar. They were about lm wide at the base and 0.30m high (Section No. 7). Their purpose must have been to infill functionally useless parts of the hypocaust system and so direct heat more efficiently to the enlarged Hot Room above. In Kent, similar wall ramps have been recorded in bath buildings at Lullingstone (Meates 1979, 97; Rook 2002, 31), Darenth (Payne 1897, 61) and Little Chart ( Eames 1957, 136). Deposits of carbon and ash occurred over the sub-floors of the hypocausted rooms (Fig. 7, Section Nos. 4 & 7, contexts 3022, 3094, 3111, 3112, 3113, 3114 & 3115). These were clearly derived from the workings of the heating system. Layer 3094 produced thirty-seven 42 MINSTER-IN-THANET ROMAN VIl...LA: INTRODUCTION AND BATH-HOUSE pot-sherds datable to the mid-late second century AD. Similar ashy deposits (Section No. 7, contexts 3051, 3125 & 3158) across the floor of the furnace room (Room 24) contained a further forty-one sherds of later second-century date, with a few pieces that could continue into the earlier third century. These layers also contained fragments of painted plaster and pieces of tesserae, all presumably derived from the interior decoration of the baths. In the southern half of Room 24 a central tile-edged channel with raised clay and rubble platforms on either side (Fig. 5, contexts 3155 & 3156), must relate to the main furnace. Quite possibly this furnace lay below a hot-water boiler that was supported on the platforms. A shallow depression with heavily burnt sides about half way along the channel appeared to represent the site of the actual fire. This hollow, probably caused by the continual burning and raking out of the ash, had subsequently been repaired, using clay (Fig. 7, Section No. 7, context 3073). A flint base set in the north-west corner of Room 24 (Fig. 5; Fig. 6, Section No. 5, context 3127) seemed to represent the base for a wooden stairway giving access from the outside. The base was partially bonded in yellow mortar of a similar type to that used in the outer drain channel of Phase 2B (see below), implying that the base was a later addition to the original structure and providing some evidence that Phases 2A and 2B might be contemporary. The rubble core of the base [3097] contained broken fragments of painted wall plaster and a few tesserae, indicating that this was all re-used material derived from elsewhere in the villa complex. A single post-hole filled with ash and carbon [F. 3154 J located towards the north-eastern corner of the room might represent some sort of temporary support for a n ageing roof timber. Phase 2B, Improved Drainage Externally, a stone wall set in yellow mortar and incorporating re-used fragments of painted wall-plaster had been added along the north [3038] and part of the east [3037] side of the original bath building, to create a sub-surface drainage channel, about 0.50m wide and at least 0.40m deep (Figs 5 & 6 and Plate II). Presumably, this never continued above the contemporary ground-surface. The channel terminated at the north-eastern corner of Room 23, connecting with the inserted latrine pit there (see above). There seems little doubt that this drain was intended to take waste away from the latrine as well as the discharge from the cold plunge-bath in Room 20. The new drain ran as far as the north-western corner of the main building, where it probably emptied into an open ditch (see below). An external buttress added onto the corner of the bath-house, was presumably intended to prevent water undermining the structure here (Fig. 5). 43 D.R.J. PERKINS It seems very probable that an open ditch, surviving as [F. 3163) outside the villa enclosure wall, had originally continued eastwards as far as Room 20 and served to conduct water away from the bath during Phase 1. The new stone drainage channel of Phase 2B had apparently been built into this existing ditch (Fig. 6, Section No. 6) but was also extended beyond the ditch, around the corner of the building, in order to connect with the new latrine in Room 23 (Fig. 5; Fig. 6, Section No. 5 and Plate II). In contrast to the original open ditch, the new channel was most probably covered with stone slabs or tiles and must have represented a major improvement to the bath-house drainage system. Phase 3, Extended Drainage Sometime after the external drain had been added in Phase 2B, the channel was further extended by almost 7m, beyond the villa boundary wall (Phase 3). This later extension was represented by two mortared flint walls [3041 & 3042), set 0.47m apart, butted onto the end of the original channel (Figs. 5 & 7 and Plate II). They stood to a height of about 0.64m (5-6 courses). As with the Phase 2B channel, it would seem that the drain extension had been built into what had originally been an open ditch. Exploratory excavation below the base of the drain walls revealed a layer of clay [2 I 6] which must represent back-filling of the earlier ditch (Fig. 7, Section No. 9). This pre-drain in-fill produced some painted plaster and tile but no pottery. Beyond the villa boundary wall, the drain remained as an open ditch heading northwards (Fig. 5; F. 3163). The shallow drainage gully [F. 3013] leading away from the west side of the baths must have originally discharged into the main ditch but in Phase 3 it was cut across by the new stone drain channel. This served to block it and must have rendered it inoperable. The final filling of the gully [217) produced tile and fourteen pot-sherds broadly datable to the second century. Abandonment Eventually the bath-house was abandoned. The drainage system became blocked with silt, building debris and domestic rubbish and the building's walls and hypocausts were robbed. The available evidence combines to suggest that the use of the building came to an end sometime in the early third century. The main house seems to have been given up at about the same time. The final filling of the latrine pit in Room 23 and adjacent stone channel [ 3072, 3121, 3122 & 3123] produced a coin of Faustina II, dated AD 145-161 (SF 533; see below) and thirty pot-sherds datable to the 44 MINSTER-IN-THANET ROMAN VIl...LA: INTRODUCTION AND BATH-HOUSE period c. AD 120-170. These included three pieces of samian ware. A quantity of tesserae were recovered from the upper filling of the latrine pit and these may be derived from the adjacent floor of Room 23. The filling of the main drainage channel (212, 213, 222, 3053, 3054 & 3085]; Fig. 6, Section Nos. 5 & 6; Fig. 7, Section No. 9) yielded a bone needle, painted wall plaster, tile, a few tesserae and seventy-one pot-sherds, all late second-century in date. The pottery includes three stamped samian vessels (see below, Assemblage B ). The soil and rubble filling the unlined ditch outside the villa enclosure, to the north of the stone channel [F. 3163; contexts 243-24 7] produced over I 20 sherds of pottery datable to the second half of the second century and a quantity of animal bone. The stratigraphic evidence indicates that robbing of the bath-house occurred in two successive phases. The earliest of these is likely to have been in the Roman period. At the southern end a deep robber trench (Fig. 7, Section No. 7, F. 1020/3084) had cut through earlier demolition deposits to remove the main outer walls. The filling of this robber trench produced ten pot-sherds dated c. AD 170-270. Another large, probably contemporary, robber pit (filled with 3044, 3045, 3046, 3067, etc.; Fig. 6, Section Nos. 5 & 6; Fig. 7, Section No. 7) occurred at the northern end of the building, extending along the top of the drainage channel as F. 3071 [filled with 211, 214, 215 & 221 ). This extension may well have been created by the removal of the stone slabs or tiles that originally covered the drain. The filling of the main robber pit yielded some 115 pot-sherds, all datable to the second and third centuries, whilst the filling of the extension above the drainage channel yielded a further 150 sherds, the latest of which are dated c. 240-270. Nine fragments of ceramic water-pipe were also recovered from the fill over the drain. The suspended hypocaust floors in Rooms 21, 25 and 26 shou Id not have been directly affected by these late episodes of robbing, yet they did not survive, implying that they had already been removed in an earlier phase of robbing. The general absence of post-Roman material within the robber pits may be significant and could be taken as evidence that the building had been at least partially robbed in the Roman period. FINDS Coins (not illustrated) A total of 55 Roman coins has so far been found by excavation and controlled metal-detecting across the site, although few have been 45 D.R.J. PERKINS well stratified. Only one coin was recovered from the excavation of the bath-house. This came from the filling of the stone-lined drainage channel adjacent to the latrine pit in Room 23. Site code: AFM-98-3121; SF 533 Identification: Faustina II, As or Dupondius (very worn). Date: AD 145-161 Reverse: Felicitas standing left Catalogue Ref: RIC (Antoninus Pius) 1395 Pottery from the Bath-house (not illustrated) based on notes from Malcolm Lyne To date, over 11,000 sherds of Roman pottery have been recovered from the site. The bulk of this material dates from the late first to midthird century AD. Excavation of the bath-house produced a total of around 600 sherds, mainly as small groups. Most of the sherds are of second- or early third-century date. The three following assemblages from the bath-house are worthy of specific comment. Fabric codings are those used by the Canterbury Archaeological Trust (Macpherson-Grant et al. 1995). Assemblage A [From construction layer 3050 within Room 20] The eight sherds from this context are all from a single vessel and could suggest that the building was constructed as early as the third quarter of the first century:- Handmade lid-seated storage-jar in lumpy soft fabric with up-to 2.00mm orange grog and very sparse up to 2.00mm calcined flint, fired patchy black/buff/orange. Ext. rim diameter 300mm; c. AD 50-70. Assemblage B [From the fills of the Phase 3 channel, between walls 3041 and 3042 (contexts 212, 213, 216 & 222)] The various fills of this channel yielded 71 sherds ( 1813g) of pottery. The assemblage is too small for any kind of meaningful quantification but includes fragments from an everted-rim storage-jar in North Kent Shell-tempered ware fabric B6 with decorated cordon on its neck (c. 50-170); a jar and reeded-rim bowl in sandy grey Canterbury fabric R5 (c. 70-175); a poppyhead beaker of Monaghan (1987) Type 2A5 in grey Upchurch fineware (c. 150-190) and a Class 5C3 'piedish' in BB2 fabric R l 4 (c. 150/170-250). The Central Gaulish Samian includes fragments from a Dr 38 bowl (c. 140-200), a Dr 37 bowl (c. 140-200), a Dr 40 platter (c. 150-200), a Dr 36 platter (c. 120-200) and a Dr 33 cup stamped ALBVCIANI (c. 120-160). Fragments from an East Gaulish Samian Dr 31 platter stamped MARTI. VMI (c. 150- 46 MINSTER-IN-THANET ROMAN VILLA: INTRODUCTlON AND BATH-HOUSE I 60) are also present. A late second-century date seems to be indicated for this assemblage. Assemblage C [Demolition and Robbing (contexts 203, 2 I 1, 3076)] The fill of robber pit F. 3071 ( context 211) produced l 41 sherds ( 1706g) of late first- to early third-century pottery but including four fragments from an Oxfordshire Whiteware mortarium of Young Type M22 dated c. 240-300 (Young 1977): the nineteen sherds (232g) of pottery from demolition deposit [3076] include a mortarium sherd in Oxfordshire Red Colour-coated ware fabric LRIO (c. 240-400). Building Materials Significant quantities of tile, painted wall plaster and tesserae, together with pieces of window glass, ceramic water-pipe and iron nails were recovered from the excavation of the bath-house. These materials must be largely derived from the demolished superstructure of the building and will be the subject of future reports. DATING AND DISCUSSION OF BATH-HOUSE Datable material from deposits associated with the original construction of the bath-house is limited but includes some pottery of later first-/second-century date. Fragments of tile, painted plaster and tesserae contained within the same early layers must be derived from an earlier building on the site, presumably the main house (Building 1). Although the pottery would allow for an earlier date, a construction date of c. AD 100 or just before, may be provisionally suggested for the erection of the bath-house. The bulk of the pottery recovered from the building is of second-century date and without doubt, this must have been its main period of use. During this time the structure seems to have gone through at least one phase of extensive refurbishment. The available evidence indicates that the bath-house was most probably abandoned sometime in the early third century. Subsequently, it had been quite extensively robbed, with the outer walls at the south-east corner being completely removed. Frequently in Romano-British winged corridor villas of the type represented by Building 1 at Minster, an integral bath-house is present. Clearly, this was not the case here, where the bath-building formed a separate but adjacent structure. From the positioning of this bath-house in relation to the main villa house and the presence of derived building debris within its construction levels it would seem 47 D.R.J. PERKINS that Building 3 is not a primary structure in the complex but represents a subsequent addition, built (on a slightly different axis) sometime after the main house had been erected. It seems somewhat unlikely that it would have been acceptable for the occupants of a high status dwelling like Building 1 not to have possessed baths from the first. Thus, if Building 3 is not contemporary with the primary construction of the main house, the presence of an earlier detached bath-suite seems to be implied. No clear evidence for such a structure has yet been identified, unless it is represented by a single heated D-shaped room (Room 10) at the rear of the main range of Building 1 (Fig. 4 ). Although little more than the ground plan had survived, some brief consideration to the nature of the super-structure of the bath-building is required. From its relatively small size it seems likely that the entire structure was roofed as one, with two rows of parallel vaults, under a tiled roof. Such an arrangement would have allowed the vaulting to be protected from the weather and removed the problems of a valley gutter between the vaults. The walls appear to have been painted and the windows glazed. The tesserae recovered suggest plain tessellated floors rather than mosaics. The external drains associated with the building suggest that there was a regular, if not continuous, flow of waste water from the baths. Arrangements for the in-coming fresh water supply are less clear but fragments of ceramic water-pipe (see above) seem to suggest some sort of piped supply, although its origin presently remains unclear. ACKNOWLEDGEMENTS The Abbey Farm Villa excavation team wish to express their gratitude to the Kent Archaeological Society and its Fieldwork Committee for funding and administering the training excavations, and to Mr Jack Clifton for arranging access to the site. Also to the Thanet Archaeological Society whose members spent so much time in housing and processing the finds and materials. Thanks are also due to Minster Parish Council, and the staff of Minster Agricultural Museum for their kind co-operation. In addition, the Project Director (D.R.J. P.) would like to thank the Supervising Archaeologists, Tim Allen, Emma Boast, Ges Moody, and Keith Parfitt, for their support and unflagging enthusiasm during all phases of excavation and post-excavation work. Ges Moody prepared the site drawings for publication. Finally, the hard work of all the excavators must be recorded - without their efforts little could have been achieved. 48 MINSTER-IN-THANET ROMAN VILLA: INTRODUCTION AND BATH-HOUSE BIBLIOGRAPHY Eames, J., 1957, 'A Roman Bath-house at Little Chart, Kent', Archaeologia Cantiana, LXXI, 130-146. Ireland, W.H., 1828, A New and Complete History of the County of Kent (4 Vols.), Vol. l (London). Jones, H.A., 1992, 'A Survey of Roman Fragments in Churches in SE Kent', Kent Arch. Rev. l I 0, 230-235. Macpherson-Grant, N., Savage, A., Cotter J., Davey, M. and Riddler, I., 1995, Canterbury Ceramics 2. The Processing and Study of Excavated Pottery. Meates, G.W., 1979, The Lullingstone Roman Villa, Vol. 1, The Site (K.A.S). Monaghan, J., 1987, Upchurch and Thameside Roman Pottery. A ceramic typology for northern Kent, first to third centuries A.D. (B.A.R. 173 ). Payne, G., 1897, 'The Roman Villa at Darenth', Archaeologia Cantiana, XXIJ, 49-84. Perkins, D., 1996, 'The Abbey Farm Villa Training Excavation·, Archaeologia Cantiana, CXVI, 325-9. Perkins, D., 1999, 'Roman Villa Site (Abbey Farm): Minster-in-Thanet', Archaeologia Cantiana, CXIX, 393. Perkins, D., 200 I, 'The Roman Archaeology of the Isle of Thanet', A rchaeologia Cantiana, CXXI, 43-60. Rook, T., 2002, Roman Baths in Britain (Shire). Young, C.J., 1977, Oxfordshire Roman Pottery (B.A.R. 43). 49