The Roman Villa at Minster-in-Thanet

115 THE ROMAN VILLA AT MINSTER-IN-THANET. PART 3: THE CORRIDOR HOUSE, BUILDING 4 KEITH PARFITT The excavations jointly undertaken by the Kent Archaeological Society, the Thanet Archaeological Society and the Trust for Thanet Archaeology on the Roman villa complex at Minster-in-Thanet were completed in 2004 (Perkins 1996; 1999; Perkins and Parfitt 2004; Pout 2004). Details of these investigations are being published by instalment. The first report, describing the overall location of the villa and the detached bath-house (Building 3), appeared in Vol. CXXIV of Archaeologia Cantiana (Perkins 2004). The second report was concerned with coin finds from the site and this appeared last year (Holman and Parfitt 2005). Here we present our third report, describing the substantial corridor house (Building 4) located at the south-eastern corner of the site. The villa complex lies on Abbey Farm and occupies a gentle south-west facing slope at an elevation of 13-17m above OD, overlooking the modern village of Minster and the ancient Wantsum Channel, now completely silted (Perkins 2004, figs 1 and 2). NGR TR 3135 6463, centred. Immediately to the west of the site lies a small, narrow valley containing a stream, which must have provided a valuable source of freshwater throughout the centuries (Perkins 2004, fig. 3). The excavations have revealed a series of Roman structures, most occupied between the late first and early third centuries AD. The main villa buildings were set either inside or around a large rectangular walled enclosure (Perkins 2004; Fig. 1, Buildings 1-6). The principal dwelling (Building 1) consisted of a classic ‘wingedcorridor’ villa, with a small detached bath-house immediately adjacent (Perkins 2004, Building 3). Both these structures were placed within the northern (higher) end of the walled enclosure. At the downhill end, a central gateway through the south wall of the compound gave access into this enclosed area (Fig. 1). Situated immediately outside the south-east corner of the walled enclosure lay Building 4. This was a stone-built corridor house of several phases. A corresponding external structure (Building 6) stood outside the south-western corner of the enclosure, to provide architectural KEITH PARFITT 116 Fig. 1 General plan of the villa complex showing the position of Building 4 in relation to the villa enclosure wall and main house. THE ROMAN VILLA AT MINSTER-IN-THANET. PART 3: BUILDING 4 117 balance and symmetry, in Classical style (Fig. 1). Details of the complex evolution of Building 6 and the other structures mentioned will be set out in future reports. The Corridor House (Building 4) consisted of a substantial rectangular building with stone foundations, situated some 80m downhill to the south of the main villa house (Building 1). It lay immediately outside the villa enclosure wall (Fig. 1, B4) and was excavated over two seasons, in 1999 and 2001. The work was carried out under the overall direction of Dave Perkins, after the site was initially identified by the farmer, Jack Clifton, as an area where the plough regularly turned-up building debris. Excavation showed that the walls of the building had been extensively and systematically robbed, apparently during the Roman period and subsequent ploughing had caused further damage. Nothing survived above foundation level and no floors or associated occupation layers remained, apart from the sub-floor of a small hypocausted room. A few undisturbed soil layers remained but finds of significant assemblages of pottery and other artefacts were largely confined to late dumps of domestic rubbish contained within the robbed hypocaust room and a boundary ditch [F. 5014] leading away from the north-east corner of the building (see below). The first phase of the building had been erected immediately outside the pre-existing south boundary wall of the villa enclosure, separated from it by a narrow gap. In its earliest form, the structure consisted of a rectangular building, containing two large rooms (Phase 1) surrounded by a broad corridor on the west, south and east sides (Phase 1a). The small hypocausted room had subsequently been inserted into the south-west corner of this outer corridor (Phase 2). Later, the corridor was widened and extended northwards across the line of the villa boundary wall (Phase 3), perhaps now allowing slightly easier communication with Building 1. The foundation trenches of Building 4 had been cut through a subsoil layer (context 5128) that covered the natural sandy clay in this area. A dozen sherds of prehistoric and early Roman pottery were recovered from this deposit. Partially sealed below it were two ditches [Fs 5108 and 5110], which appeared to relate to a rectangular enclosure, perhaps of late Iron-Age date, set on a completely different axis to the subsequent Roman building (Fig. 2). The South Villa Enclosure Wall (5155) and Original Boundary Ditch, F. 5003 The plough-damaged remains of a shallow wall footing constructed from rammed chalk rubble (context 5155) lay immediately to the north of Building 4. This was aligned on an E-W axis and was traced for a minimum KEITH PARFITT 118 Fig. 2 Building 4, general site plan. THE ROMAN VILLA AT MINSTER-IN-THANET. PART 3: BUILDING 4 119 distance of 17m in the excavation (Fig. 2). Further sections have been traced to the north and west and there can be no doubt that collectively these relate to an outer courtyard wall, which originally enclosed all four sides of the main villa complex (Fig. 1). The present section of foundation must be associated with the south wall of this enclosure, although its junction with the eastern wall had been completely destroyed by ploughing. Further to the west it was better preserved, and was found to be about 0.63m wide and 0.22m deep. In the area of Building 4 two small projections on its southern side, some 7m apart, fairly certainly represent bases of contemporary external buttresses (Fig. 2). The available evidence suggests that the villa enclosure wall was already in existence when Building 4 was erected; specifically, the west wall of Building 4’s Central Range is aligned on one of the boundary wall buttresses and the two successive west corridor walls (Phases 1a and 3) both seem to respect the boundary wall foundation (see below). Extending eastwards from the presumed site of the destroyed southeast corner of the villa enclosure wall was a substantial ditch (Fig. 2). In its earliest form [F. 5003] this was probably contemporary with the boundary wall and seems to have continued the line of the south wall. The ditch was about 2.40m wide and 1.00m deep with sloping sides and a dished base (Fig. 3). It perhaps helped delimit a field or enclosure beyond Fig. 3 Section across boundary ditch [F. 5003] and its re-cut [F. 5014]. KEITH PARFITT 120 the main walled compound. The construction of Building 4 eventually required the original terminal of this ditch to be back-filled with clay to make-way for a new eastern corridor wall (see below for details). Detailed Description of Building 4 (Fig. 2 and Plate I) Building 4 was aligned N-S, following the slope of the hill and lay between the 14 and 15 metre contours (Perkins 2004, fig. 3). It was apparent that several phases of development must be represented within the structure overall but with very few useful stratigraphic relationships preserved, only a tentative sequence can be suggested here, largely based on a visual interpretation of the positioning and layout of the foundations (Phases 1, 1a, 2 and 3). Phase 1, The Central Range (Fig. 2 and Plate I) At the heart of Building 4 lay a central range measuring internally some 12m (N-S) by 6.00 m (E-W). A cross-wall had originally divided this range into two rooms, of almost equal size: the North Room measured 5.98 m (E-W) by 5.85 m (N-S), whilst the South Room was 6.00 m (E-W) by 5.50 m (N-S). There seems little doubt that this Central Range represents the earliest part of the High level view of Building 4 PLATE I THE ROMAN VILLA AT MINSTER-IN-THANET. PART 3: BUILDING 4 121 building and that it constituted a modest detached dwelling, set just outside the south-east corner of the walled villa enclosure (Fig. 1). The foundations of the Central Range had been heavily robbed and the footing for the cross-wall dividing the two rooms had been completely removed. More than half the north wall foundation had also been removed (Fig. 2 and Plate I). In most places, however, at least the basal course of foundation stones remained. Where surviving, the foundations for the walls were of similar dimensions and construction. They ranged in width from 0.75-0.80m, with a maximum remaining depth of 0.35m. The original depth of these foundations must have been at least 0.40m. The robber-trench for the foundation of the cross-wall between the North and South Rooms was about 0.60m wide and 0.28m deep, indicating that this internal wall had, typically, been of less substantial construction than the outer walls. From what remained, it could be seen that the foundation stones had originally been carefully laid in their construction trench. Large flint beach cobbles were placed along the outer edges with courses of smaller cobbles used as a central in-fill. The cobbles were set in a matrix of light brown gritty loam which detailed inspection showed to contain significant quantities of crushed mortar and plaster, including numerous small fragments with a painted surface. There seems little doubt that this is re-used demolition material derived from another structure on the site. Examination of the associated flint cobbles failed to reveal any with traces of adhering mortar but this need not preclude at least some of these also being re-used. Whether the walls of the Central Range above the robbed foundations were wholly or only partially built of stone remains uncertain. The footings would certainly appear to have been substantial enough to have supported an entirely stone-built structure which could have included an upper storey. The surviving foundations were sealed by the back-filling of their respective robber-trenches, which largely consisted of soil, Roman tile fragments and loose flint cobbles (generally the smaller sizes), with much crushed mortar and painted plaster. The painted plaster is presumably a mixture, comprising some early fragments originally incorporated into the foundation matrix with most other pieces derived from the demolished walls and ceilings of Building 4 itself. The exact date at which the robbing of the walls took place remains unclear, although the probability is that it was sometime during the late Roman period on the evidence of a few late third/fourth-century potsherds recovered from the robber-trench fills (see below). Certainly, no post-Roman finds came from the back-filling of any of the trenches. A group of four Roman post-holes was located within the Central Range (Fig. 2). Three of these features occurred in the South Room [Fs KEITH PARFITT 122 5136, 5138 and 5140] with one in the North Room [F. 5142]. Of these, Fs 5136, 5138 and 5142 fell along the centre-line of the building, suggesting that they could be related to some sort of contemporary internal wooden partition, which perhaps sub-divided each of the main rooms. Alternatively, they perhaps related to posts added to support the floor joists of an upper storey – packing stones of re-used flint cobbles and tile could suggest that these post-holes were late inserts. Two other postholes [Fs 5146 and 5153] and a small pit [F. 5151] appeared to be earlier and unrelated to the building. Post-hole F. 5146 had been cut through by the foundation trench for the east wall of the Central Range but failed to yield any dating evidence. A spread of demolition rubble (5127), largely undisturbed by ploughing, had survived in the north-eastern quarter of the North Room (Fig. 2). This deposit was up to 0.15m thick and produced three sherds of a Roman jar, which are not closely datable, a quantity of Roman tile and a piece of whale bone. No floor levels were preserved below it, which might suggest that the original floor of the room had been of timber and that it had been removed at an early stage during the building’s demolition. Phase 1a, The Outer Corridor Wall (Fig. 2) The Central Range was surrounded on the west, south and east sides by a broad corridor or gallery, 2.14-2.33m in width internally (Fig. 2). Its outer walls had again been thoroughly robbed and there is no surviving evidence to show whether the walls were exactly contemporary with the Central Range or represented a later addition. The differing width, depth and construction of the corridor foundations in comparison with those of the Central Range could be taken to indicate that they were a subsequent addition but this variation might equally well reflect nothing more than the materials available and the respective wall loadings intended. In view of the uncertainty of its dating, the corridor surrounding the Central Range has been designated here Phase 1a. In terms of construction, the corridor foundations were similar on all three sides, consisting of a compact, roughly coursed mixture of flint nodules and cobbles with some tufa, sandstone, chalk, mortar lumps and pieces of Roman tile, all set in a brown gritty loam with some small mortar fragments. A number of flints showed clear evidence of adhering mortar and there can be no doubt that the foundations, as with the Central Range, were at least partially formed from re-used material. The construction trenches for the corridor foundations were shallower than those dug for the walls of the Central Range. Where surviving, the corridor foundations were 0.65-0.70m across and 0.10-0.25m deep. In most places, only the bottom course of stones remained but two courses survived at the south-east corner. The northern end of the west foundation THE ROMAN VILLA AT MINSTER-IN-THANET. PART 3: BUILDING 4 123 stopped with a neatly squared terminal 0.18m short of the chalk footing of the south villa enclosure wall (5155, see above). From this, there seems little doubt that the enclosure wall was already in place when the corridor wall was constructed. The new wall must have been keyed into this existing boundary wall at a higher level, so that the boundary wall itself would have then become the corresponding corridor wall for the north side of the building, even though the passage here was much narrower than on the other sides (about one metre). At the northern end of the east wall the foundation had been wholly destroyed by ploughing but there seems little doubt that it originally abutted the south-eastern corner of the villa enclosure wall, crossing the line of the adjacent boundary ditch, F. 5003. The end of this ditch must have been deliberately in-filled to accommodate the new wall. A roughly square mortar base, about one metre across, lay over the filled-in ditch terminal here (Fig. 2). This might represent the base of the corridor wall where it connected with the enclosure wall. The soil and rubble filling the robber-trenches associated with the Phase 1a corridor walls yielded a few sherds of Roman pottery, datable to the second/third centuries, together with fragments of painted plaster and Roman tile. Whilst the pottery might be largely residual, there can be no doubt that the walls of this original corridor had been removed later during the Roman period in order to make way for a new, wider corridor (designated Phase 3; see below). Phase 2, The Inserted Hypocaust Sometime after its original construction, a short section of the Phase 1a corridor wall had been removed in order to allow the excavation of a 0.50m deep, L-shaped construction pit for the insertion of a small hypocausted room at the south-west corner of the building (Figs 4 and 5). This new room was placed in the south corridor, with the associated stoking chamber at the end of the west corridor. As with the rest of the building, this room had subsequently been extensively robbed but sufficient evidence remained to allow the basic arrangement of the pillared hypocaust system to be reasonably clear (Fig. 4). The room had internal dimensions of 3.20m (E-W) by 1.98 m (N-S). Its sub-floor (5117) consisted of a layer of coarse cream mortar poured onto flint cobbles with a total thickness of 0.16m. This sub-floor supported the outer walls and a network of central pilae. The walls had been mostly robbed but the lowest courses survived at the corners. They consisted of flint cobbles set in a hard cream-white gritty mortar and survived to a maximum height of 0.20m. They were 0.19-0.26m wide, the south wall being re-instated on exactly the same line as the original corridor wall (Fig. 4). KEITH PARFITT 124 Fig. 4 Detailed plan of hypocausted room. THE ROMAN VILLA AT MINSTER-IN-THANET. PART 3: BUILDING 4 125 Fig. 5 Section across hypocausted room. KEITH PARFITT 126 Some twenty-six pilae stacks had once supported the upper floor of the heated room. Remains of only four base tiles survived the later robbing but scars of most of the other stacks could be discerned on the surface of the mortar sub-floor (Fig. 4). It was apparent that some of the tiles had been fixed to the floor by a deposit of brown clay but others were set on a pad of white mortar. This seems to imply that the hypocaust was in use for a significant length of time and that a number of the stacks had required repair or rebuilding at some stage. The small, rectangular stoking-chamber lay at the north-west corner of the main room, although very little of its structure had survived. An area of poured opus signinum over flint cobbles, measuring 0.75m (E-W) by 0.58m (N-S), abutted the northern edge of the main sub-floor and this is likely to represent the base for the furnace arch. The stoking-chamber itself seems to have been unlined except for a short length of mortared flint cobble walling on the west side which might conceivably have served as a stairway support, although access can only have been from the north (Fig. 4). Clearly, some major above-ground alterations would have been required to allow the insertion of this new enclosed structure and it seems unlikely that there could now have been any through-access between the west and south corridors, which was perhaps somewhat inconvenient for the building’s occupants. The positioning of the stoking chamber within the west corridor, rather than outside the building, tends to suggest that the corridor on this side was an open colonnade, rather than an enclosed gallery, thereby allowing smoke and fumes from the furnace to easily disperse and also providing good air circulation for the working of the system. The exact function of the hypocausted room remains uncertain. It could be regarded as a heated living room but its small size and peripheral location within the outer corridor rather than in one of the main central rooms of the building, seems to argue against this. Black (1987, 49) has described how small heated rooms were arranged in some early Roman villa houses around a main, unheated room, so as to provide indirect heat through the walls. Such a primary function for the present room, however, seems unlikely not least because of the lack of other similarly heated rooms placed to serve the other sides of the main room (which here would have been the South Room of the Central Range). Nevertheless, some indirect heat would probably have passed through to the adjacent South Room as a useful by-product. Perhaps the most attractive idea is that this well-built, heated room represents a small office or private chamber – a function possibly reflected in its location away from the main living rooms (see below). An alternative, if less likely interpretation, may be that the structure served as a rather elaborate drying room used for processing grain or other agricultural produce. Certainly, there are many examples of various drying kilns occurring within villa houses. THE ROMAN VILLA AT MINSTER-IN-THANET. PART 3: BUILDING 4 127 Eventually, the heated room was abandoned and the structure was systematically robbed down to basement level (Fig. 5). There can be no doubt that this occurred sometime during the Roman period as large amounts of domestic rubbish, including some 1,700 sherds of pottery, were subsequently dumped into the basement. The date of the pottery spans the period c.100-250. This could indicate that the material had been dumped over a considerable period of time (Lyne 2002) but more probably it reflects the fact that old midden material had been deliberately brought in from elsewhere. What is less clear is the exact time of the robbing in relation to the overall structural history of the building. The walls of the Phase 1a corridor had clearly been removed at the start of Phase 3 to allow the enlargement of the corridor (see below). Thus, the removal of the hypocausted room might also have occurred at this time; however, since no new heated room was provided as a replacement, this might appear as a retrograde step; so perhaps the existing hypocausted structure was incorporated into the later corridor, albeit rather awkwardly. If this was the case, the robbing of the hypocaust must have occurred later, presumably when the rest of the building was robbed, after Phase 3. In view of the presence of some third-century material within the pottery assemblage this perhaps seems most likely. The evidence of the re-built pilae could also suggest a fairly long period of use for the hypocaust. Furthermore, had the heated room been removed at the start of Phase 3 to make way for the new corridor, a solid infill, such as clean clay, sand or rubble might have been most appropriate in order to support the overlying floor of the new passage. Unconsolidated domestic refuse would almost certainly have been unstable and led to subsidence of the new floor into the earlier basement but no evidence of such an occurrence was observed during the excavation. Whatever the precise date of the robbing of the hypocausted room, the pottery and other domestic rubbish dumped into its basement was clearly intended to in-fill a large, unwanted hole. Sometime after the heated basement had been robbed, back-filled and levelled, an oval pit was partially cut through its northern end (Fig. 4, F. 5112). This pit also cut into the foundation of the Central Range and must be a late feature. Its filling produced sixty-eight pot-sherds, all datable to the later second-early third century, but these are clearly derived from the preceding basement infill deposits and cannot be used to provide any close dating for the filling of the pit itself, which could be post-Roman. Phase 3, The New Outer Corridor Wall (Fig 2 and Plate I) Shallow trenches located about 0.50m beyond both the east and the west corridor wall foundations assigned here to Phase 1a indicate these walls had subsequently been replaced by new ones, to provide a broader corridor, now some 3.50m in width. All trace of any corresponding KEITH PARFITT 128 foundation trench on the south side had been destroyed by the plough but on the north side another shallow trench indicated that a corridor of similar width had now also been created along this side of the Central Range. The south wall of the villa enclosure must have been removed to accommodate this new corridor (see below). The construction of a new corridor wall at the north (and presumably the south) represented a significant increase in the length of Building 4 (Fig. 2). The overall dimensions for the structure during Phase 3 were probably about 22m (N-S) by 16m (E-W). The foundation trenches for the new western corridor wall stopped short of the pre-existing south villa enclosure wall on both sides (Fig. 2). This can be explained if the boundary wall west of the line of the new corridor wall had been retained, with everything to the east removed to foundation level in order to allow free access into the new north passage. The new corridor walls could then have been carefully stitched into the stub of the remaining enclosure wall. Such a sequence may be reflected by the better preservation of the boundary wall foundation west of the new corridor. The Phase 3 corridor walls had, themselves, subsequently been robbed, presumably at the same time as the walls of the Central Range were removed. The corridor robber-trenches varied from 0.60-0.75m in width and were 0.05-0.15m deep. All the associated stonework had been taken so that the nature of the foundations they originally contained remains unknown. From the size of the robber-trenches, however, a footing of similar dimensions to the Phase 1a corridor wall is implied. Indeed, it seems highly likely that the earlier corridor walls provided much of the building material used for the Phase 3 works. The soil and rubble filling the robber-trenches associated with the Phase 3 walls yielded half-a-dozen sherds of Roman pottery broadly datable to the second/third centuries, together with fragments of painted plaster, Roman tile, animal bone and oyster shell. An un-robbed length of wall footing (Fig. 2, 5147) linking the line of the Phase 3 west corridor wall to the Central Range could be contemporary with the new corridor or a subsequent addition. This surviving foundation consisted of one-two rough courses of large flint cobbles 0.44-0.55m in width and 0.15m deep, set in brown clay loam. A number of the flints showed evidence of adhering mortar indicating that, again, they were re-used. The foundation could be clearly seen to cut through the robbertrench associated with the Phase 1a corridor wall, which confirms that this had been previously removed. Traces of a very shallow robber-trench running from the line of the Phase 3 corridor wall on the east side seem to mark the position of a corresponding cross-wall almost directly opposite. It would thus appear that the Phase 3 corridor had once been partially subdivided. Other cross-walls could have been completely destroyed THE ROMAN VILLA AT MINSTER-IN-THANET. PART 3: BUILDING 4 129 and it may be that the later corridor walls in reality defined ranges of narrow rooms enclosing the Central Range. Similar sub-division of the outer corridors occurred in the later phases of the main villa house on the present site (Building 1) and is also known at a number of other villa sites, e.g. the South Building at Keston, in west Kent (Philp et al. 1991). All trace of the northern part of the new eastern corridor wall had been destroyed by ploughing but a section of deeper foundation survived where it had been necessary for the builders to dig into the soft backfilling of the earlier boundary ditch [F. 5003], see above. The 0.50m wide foundation here (5022) had been built free-standing in a wide construction trench and consisted of roughly coursed, large (re-used) flint cobbles set in dark brown gritty loam, with traces of cream-yellow mortar bonding in the upper course. As elsewhere, this foundation had been extensively robbed but sufficient remained to suggest that it had originally been cut some 0.55m deep into the ditch fill, of which the lowest 0.25m of stonework remained. A few sherds of second-century pottery had become incorporated into the fabric of the foundation. Immediately to the east, the line of the original boundary ditch [F. 5003] was maintained and recut as F. 5014, with its repositioned terminal placed just outside the Phase 3 corridor wall, about 2.40m east of the original terminal (Figs 2 and 3). At the end of its life the re-cut ditch had been backfilled with large amounts of roof-tile, pottery and other domestic rubbish. The pottery all falls within the range c.170-250 which suggests that the ditch was finally filled around the middle of the third century. It seems most likely that this event occurred at about the same time as Building 4 was abandoned. Much of the pottery and other domestic rubbish contained within the ditch is probably derived from the building itself and may well have been re-deposited in the ditch during the final levelling and tidying of the area, after the old building had been systematically dismantled. THE FINDS No coins were discovered during the excavation of Building 4 (Holman and Parfitt 2005) but the work did produce around 4,500 pot-sherds (Lyne 2002). There are two principal pottery groups – from the infill of the robbed hypocausted room at the south-west corner of the building (1,713 sherds) and from the filling of the re-cut ditch [F. 5014] at the north-east corner (2,256 sherds). Collectively, these two assemblages account for well over three-quarters of the total pottery recovered. Smaller groups came from the filling of the original north-east boundary ditch F. 5003 (193 sherds) and the later robber-trenches associated with the building (72 sherds). Another group of mixed, unstratified material came from the plough soil overlying the building (about 170 sherds). The bulk of the pottery recovered dates to between the late first and the mid-third century, KEITH PARFITT 130 with only a few pieces of the later Roman period. Detailed consideration of these assemblages will appear in a future report. Other finds from the area of Building 4 included roof tile, painted wall plaster, tesserae, a range of small finds and food debris in the form of marine shell (mostly oysters) and animal bone. Again, these items will be reported upon later. DATING AND GENERAL DISCUSSION The general lack of occupation layers and other stratified deposits preserved within the structure limits any attempt to closely date the use of Building 4. In the absence of coinage, the pottery provides the best guide. The few early Roman sherds recovered from the subsoil layer (5128) predating Building 4 imply that it cannot have been erected before the last quarter of the first century AD at the earliest, whilst the presence of the pre-existing villa boundary wall suggests that Building 4 is rather later than this and fairly certainly it did not form part of the primary layout of the villa complex. Other evidence helps to confirm this; the presence of re-used material within the Phase 1 fabric of Building 4 indicates that this cannot have been the first masonry building erected on the site and the occurrence of derived fragments of painted wall plaster within the make-up of the primary foundations implies that another significant structure, presumably the main villa house (Building 1), had already been in existence for a sufficiently long period of time to have undergone redecoration and/or alteration, sometime before the initial construction of Building 4. In the area of Building 4 itself, the terminal of the original north-east boundary ditch [F. 5003] seems to have been deliberately backfilled to allow the erection of a new eastern corridor wall during Phase 3 of the building, if not before. The pottery from the ditch fill ranges in date from c.70-170 and from this it seems unlikely that the Phase 3 corridor of Building 4 could have been laid out before the late second century. The large pottery assemblages recovered from the in-filling of the re-cut boundary ditch [F. 5014] and the robbed hypocaust room at the southwestern corner of the building both fall within the period c.100-250 and this appears to encompass the maximum date-range for the use of Building 4, even if these groups of material were not deposited in their excavated locations until the final abandonment of the structure (see above). Most probably, Building 4 was first constructed soon after the middle of the second century. The general absence of late third and fourth century Roman pottery across this part of the site is quite noticeable and further reinforces the view that Building 4 had been abandoned by the middle of the third century. Subsequently, however, extensive stone robbing occurred here. The robber-trenches associated with the building produced about 70 pot THE ROMAN VILLA AT MINSTER-IN-THANET. PART 3: BUILDING 4 131 sherds (see above). The bulk of these are of late first- and second-century date, with just two or three pieces that date to the late third or fourth century. It seems most likely that these few late Roman sherds reflect the date of the main robbing of the structure, with the bulk of the pottery representing residual material derived from the original occupation of Building 4. The general layout of Building 4, with its two-roomed Central Range surrounded by broad corridors, a hypocausted room, evidence for painted walls and probable tessellated floors, leaves no doubt that this structure represents a domestic dwelling, rather than an agricultural building. In comparison with the main villa house (Building 1), however, it was a rather smaller structure, even at its maximum extent during Phase 3. In overall plan, Minster Building 4 is broadly comparable to the South Masonry Building at Keston (Philp et al. 1991, 120-5) and also Block A of the villa complex at South Darenth (Black 1981, fig. 2), both in west Kent. The Keston building was in use during the late third and fourth centuries, later than Building 4. One difference in plan between the two west Kent corridor houses and Building 4, however, is Minster’s absence of any cross-passage separating the two central rooms. As at Abbey Farm, however, the two west Kent houses formed elements of more extensive villa complexes and both seem to represent dwellings that were ancillary to the main house. The provision of broad corridors around these buildings may also be matched by the later arrangements in the main villa house at Minster itself (Building 1). The presence of such broad corridors or galleries could reflect Gallic influence in the design of these structures (Black 1987, 140). As at Keston and Minster Building 1, there is some evidence to suggest that at least certain sections of the corridors around Building 4 had once been sub-divided by cross walls to create a number of outer rooms, one of which was hypocausted. The existence of broad, roofed outer corridors, eventually around all four sides of Building 4, would have had some bearing on the lighting of the Central Range. If the core of the building was two storeys high, windows set above the corridor roof lines could have admitted direct light into the middle of the structure, although this would have been more limited if the building was provided with a continuous upper floor. At ground-level, more light would have been admitted if the corridors were open colonnades rather than enclosed galleries. Yet the evidence of cross walls in these corridors, together with the hypocausted room, implies that at least some enclosed apartments existed. The positioning of Minster Building 4 in relation to the main villa house appears to be of some significance and may well be a reflection of differences in the social status of their respective inhabitants. Building 4 was set at a lower elevation and placed outside the walled enclosure surrounding the main villa complex, well away from the principal house. KEITH PARFITT 132 Such details seem to imply that there were no close familial links between the occupants of the two residences. Although Building 4 would appear to have been a reasonably comfortable dwelling, its smaller size in relation to Building 1, together with the use of significant amounts of re-cycled material in its foundations, might be taken to suggest this was a building for which moderate construction costs were required. Accepting that the small, isolated heated room at the south-western corner does not relate to such a structure, the absence of any integral or immediately adjacent bath-suite further implies that the occupants of Building 4 were not of the highest status. Nevertheless, a bath-suite, quite separate from that adjoining Building 1 and perhaps specifically intended for use by those who lived outside the main villa enclosure, was provided at one stage within Building 6 (A), the matching block to Building 4, located at the opposite corner of the villa enclosure (Fig. 1). The overall impression gained is that Building 4 was a dwelling for the use of inhabitants who were of lower status to those living in the main villa house, although still closely connected with the establishment. Such could have been the head bailiff (rusticus) of the villa estate and his family. Presumably a man of some importance locally, the estate bailiff could have been provided with this comparatively modest but comfortable dwelling, placed at a discreet distance from the main villa house. Such a positioning would seem entirely appropriate for an employee, who nevertheless would have remained near enough to be easily contacted and well-placed to receive instructions from the estate-owner up at the main house. In this context the detached south-west heated room in Building 4 may become of particular significance and might be interpreted as the estate office where the rusticus worked on the farm accounts and where routine business could be transacted. Easy access to this room for visiting tenants and other outsiders may well explain its location in an outer corridor, rather than within the central part of the house. The available information suggests that Building 4 constituted a subsequent addition to the original layout of the Minster villa complex. Such a development could be interpreted as providing evidence for the growing prosperity and status of the proprietor of the estate, who perhaps increasingly distanced himself from the day-to-day running of a working farm. Eventually, such routine matters were delegated to a trusted bailiff, who was provided with his own house outside the owner’s private walled enclosure. ACKNOWLEDGEMENTS The Abbey Farm Villa excavation team wish to express their gratitude to the Kent Archaeological Society for funding and administrating the excavations. In particular, the then President, Paul Oldham must be THE ROMAN VILLA AT MINSTER-IN-THANET. PART 3: BUILDING 4 133 warmly thanked, together with David Bacchus and Chris Pout of the Fieldwork Committee. Farmer Mr Jack Clifton again arranged access to the site and has taken a keen interest in the work throughout. Members of Thanet Archaeological Society have spent much time 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 Directors (Dave Perkins and Keith Parfitt) would like to thank the Supervising Archaeologists, Tim Allen, Emma Boast and Ges Moody, for their support and enthusiasm during all phases of excavation and post-excavation work. Ges Moody and Susan Deacon prepared the site drawings for publication. Finally, the hard work of all the many excavators must be recorded – as always, without their efforts little would have been achieved. BIBLIOGRAPHY Black, E.W., 1981, ‘The Roman villa at Darenth’, Archaeologia Cantiana, XCVII, 159-183. Black, E.W., 1987, The Roman Villas of South-East England (B.A.R. British Series 171). Holman, D.J. and Parfitt, K., 2005, ‘The Roman villa at Minster-in-Thanet. Part 2: the Iron-Age, Roman and Later Coinage’, Archaeologia Cantiana, CXXV, 203-228. Lyne, M., 2002, The Pottery from the Minster Roman Villa (AFM 1996-2002) (archive report commissioned by the KAS). 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., 2001, ‘The Roman Archaeology of the Isle of Thanet’, Archaeologia Cantiana, CXXI, 43-60. Perkins, D., 2004, ‘The Roman Villa at Minster-in-Thanet. Part 1: Introduction and Report on the Bath-house’, Archaeologia Cantiana, CXXIV, 25-49. Perkins, D. and Parfitt, K., 2004, ‘The Minster Roman Villa’, Current Archaeology, 193, 38-41. Philp, B., Parfitt, K., Willson, J., Dutto, M., and Williams, W., 1991, The Roman Villa Site at Keston, Kent First Report (Excavations 1968-78) (KARU, Dover). Pout, C., 2004, ‘Final Excavation at Abbey Farm’, KAS Newsletter 62, 1-2.