Preliminary Report on Excavations at Hull Place Roman Villa Sholden 2005-7

103 PRELIMINARY REPORT ON EXCAVATIONS AT HULL PLACE ROMAN VILLA, SHOLDEN, 2005-7 keith parfitt The existence of a probable Roman villa close to Hull Place at Sholden, near Deal, has been known since 1921 when the outlines of a long-demolished building were revealed as a parch-mark. An account of the original discovery was given by Henry Chapman in his local history study, The Story of Dola (Chapman 1921, 174) but very few further details have been available (Phillips 1964; Parfitt 1980; 1986). Some excavation was undertaken, soon after the initial discovery, by the landowner Gilbert Elliot and local archaeologist William Stebbing, both KAS members (Igglesden 1935, 61-2; Stebbing 1937, 19, 41). No detailed report on their work was ever published and by the start of the twenty-first century the precise location and status of this potentially important Roman site were largely unknown. Research into the evolution of the Lydden Valley marshes, with the suggestion that the reclamation of these might have begun as early as the Roman period (Lydden Valley Research Group [LVRG] 2006, 35-6), re-focussed attention on the role of the presumed villa at Hull Place, situated immediately adjacent to the marshland (Fig. 1). This, together with the continued development of the area occupied by the villa as a popular fishery and leisure park (Cottington Lakes), indicated that a re-examination of this little known Roman site was becoming increasingly important, both from the point of view of local archaeological research and conservation management. Accordingly, a programme of investigation was undertaken by members of the Dover Archaeological Group between 2005 and 2007, working in association with the newly formed LVRG. The primary aim of the project was to re-confirm the existence of Roman building remains here and to establish something of their preservation, character and date-range. Particular importance was attached to determining the extent of the remains so that appropriate sections could be left undisturbed in any future development of the area. Working with a small team on a part-time basis, in a field used as a camping ground, total KEITH PARFITT 104 Fig. 1 Map showing location of the Hull Place, Sholden villa in relation to local topography and other Roman sites. excavation of the complete complex was not feasible; nevertheless, much valuable new information has been recovered. This has greatly enhanced our understanding of the site and has clearly demonstrated the presence here of two separate, apparently successive villa houses (designated Buildings A and B, see below), sealing evidence for preceding occupation EXCAVATIONS AT HULL PLACE ROMAN VILLA, SHOLDEN, 2005-7 105 during prehistoric and earlier Roman times. The following account is intended as a preliminary description of the findings; a more detailed report will appear in due course. The villa site lies some 830m to the north of Sholden’s medieval parish church and about 200m north-east of Hull Place, a seventeenth-century brick-built mansion. It is located at the foot of Sholden Downs in a field of mown grass (O.S. Parcel No. 7200) now used by touring caravans but once part of the grounds of Hull Place. NGR TR 3572 5297, centred. The surface of the field slopes gently down to marsh-level, with the site of the villa standing at an elevation of between just 2.20 and 1.30 m. above od. The natural subsoil here is head brickearth supporting topsoil belonging to the Hook Series (Fordham and Green 1973, 59-62). Beyond the site, wide tracts of coastal marshland relating to the Lydden Valley extend away to the north and east (Fig. 1), although landscaping associated with the Cottington Lakes complex and the former Betteshanger Colliery coal-tip now partially obscure the true character of the region. As mapped by the Geological Survey, marsh alluvium begins around 100m from the north-eastern end of the excavated Roman buildings. At the start of the excavations in 2005 the supposed site of the Roman structure was visible as a slight mound in the central part of the field. Inspection showed this mound to be roughly square in shape, about 15m across. A geophysical survey of the surrounding area suggested the presence of several buried walls but, disappointingly, no coherent overall plan was revealed. This initial survey work was followed by excavation. Roman, flint-built foundations were soon located, with the highest fragments buried at a depth of only 0.20m but the mound visible on the surface proved to be composed of modern soil and rubble infilling the excavation of the 1920s (Room 24; see below, Plate I), rather than undisturbed remains of a collapsed Roman building. Pre-Villa Occupation Below the Roman buildings, a number of earlier features, including pits, post-holes, ditches, gullies and ovens, were located. These remains extended across a considerable area, although too little was seen to discern any coherent overall layout. Nevertheless, they clearly represent pre-villa activity on the site and variously produced pottery and other finds of late Iron Age and early Roman date. Several separate phases must exist. Of some particular interest was a group of three successive, sub-surface, domestic ovens, sealed below the eastern corner of Building A. The latest of these was remarkable in having part of its domed clay roof still surviving intact. Finely laminated clay sediments partially infilling the structure appeared to have been deposited by water and it would KEITH PARFITT 106 seem that this oven was probably abandoned due to flooding. The small assemblage of pottery associated with these ovens suggests that they were in use c. ad 40-60. Several other early hearths/ovens were located in other parts of the excavated area. The line of a substantial early boundary fence, running north-east by south-west across the excavated area to the north-west of Building B, was represented by a row of flint-packed post-pits. This fence-line had subsequently been replaced by a buttressed wall following the same course. The new boundary wall had foundations of rammed chalk and was traced for a total distance of some 45m across the site. The replacement of the earlier fence-line with a continuous wall may be connected with the development of Building B, the first villa house, and perhaps helped to delimit an associated compound. From the remains recorded it can now be established that the Hull Place villa complex, like many others in Kent and beyond, had evolved from an earlier farmstead site established before the Roman Conquest. Associated with the earliest occupation are two Celtic coins. A light scatter of prehistoric struck flints demonstrates even earlier occupation on the site, during the Neolithic/Bronze Age, whilst the discovery of a locally rare microlith of Mesolithic date hints at activity in the area prior to this. Hypocausted Room 24, Building B, previously excavated in 1921, looking west PLATE I EXCAVATIONS AT HULL PLACE ROMAN VILLA, SHOLDEN, 2005-7 107 The Roman Villa Complex The existence of a substantial Roman villa complex at Hull Place is now fully confirmed. The investigations have established that building remains extend across an area almost 40m square, with further associated features and deposits continuing for some distance beyond this (for outlying features to the south, see Parfitt 1992). The size of the area covered by the excavated building remains equates reasonably closely with Chapman’s roughly paced dimensions, given as 30 by 50 yards (Chapman 1921, 174; see above). The foundations revealed in 2005-7 clearly related to two separate Roman structures, designated here Buildings A and B. The general ground-plan of both was ascertained, although the complete outline of neither could be fully exposed. The state of preservation of both Roman buildings was found to be poor, with no floors and virtually nothing structural surviving above foundation level. Extensive lengths of walling in Building A had been totally robbed away during the medieval period, and what remained had subsequently been truncated. It seems likely that the entire area has been ploughed in past centuries and during the early twentieth century Gilbert Elliot created around Hull Place ‘a magnificent garden and acquired land stretching almost to the sea’ (Igglesden 1935, 61). This implies that some extensive landscaping occurred and there was evidence to suggest that the soil covering the villa site had been turned-over and reduced as part of this process. Building B, buried at a slightly greater depth, had been less extensively damaged by post-Roman activity than Building A but its walls appeared to have been systematically removed during the Roman period, leaving only its foundations. The two Roman structures excavated seem to represent successive dwellings, the earliest being Building B. This structure was probably first erected in the early second century and went through several phases of development before being superseded by the larger Building A, positioned on a new site immediately to the north-west. At the end of its life, sometime in the third century, the walls of Building B seem to have been dismantled, most probably in order to provide raw materials for the new Roman structure, Building A. The walls of Building A had been robbed in medieval times, implying that its ruins were still visible when they attracted the attention of twelfth/fourteenth-century workmen in need of a local stone supply. These medieval builders were perhaps engaged in work on the parish church or in constructing the manorial chapel at Cottington, some 800m to the north-west of the site. Building B Building B lay immediately to the south-east of Building A. It was the smaller structure and contained 11 rooms and corridors. Excavation was KEITH PARFITT 108 able to recover a largely complete ground plan, although the building could not be fully exposed. From the surviving flint foundations, it was possible to identify five separate phases of development and in its final form the building consisted of a block measuring a maximum of 23.50m (ne-sw) by 15m (nw-se). Originally, Building B had consisted of a simple rectangular, single-roomed structure, with an internal length of about 11m (ne-sw) and a width of around 4.75m (nw-se). On the strength of contemporary external buttresses and an apparent lack of any internal sub-divisions, it seems possible that this earliest building might have been an agricultural barn, rather than being domestic in function. If so, it probably formed part of the native farmstead that preceded the main villa complex, other evidence for which was also recorded (see above). The subsequent addition of further rooms and corridors significantly enlarged the original structure and these were clearly associated with domestic use by individuals of some social standing. The new rooms included a large one projecting from the middle of the north-west side of the original building (Room 18) and another with a hypocaust at the west corner (Room 24; Plate I). Later, a room with an apse was added to the south-western end of the structure (Room 19; Plate II). Such architectural PLATE II Apsidal room at south-west end of Building B (Room 19), looking west EXCAVATIONS AT HULL PLACE ROMAN VILLA, SHOLDEN, 2005-7 109 refinements, together with evidence for the use of painted wall-plaster, indicate that the building had evolved into a structure of some quality. Following its abandonment and levelling, Building B was cut through by a series of late Roman ditches and gullies, containing fourth-century coins and pottery. The very latest pottery in these is likely to date to the early fifth century. From the modern backfill contained within Room 24 (Plate I), there can be no doubt that this hypocausted wing-room was the part of the villa complex previously excavated by Elliot and Stebbing (see above). In the absence of any other evidence for recent digging, it would seem that their excavations had been confined to this one small area. Building A A significant proportion of Building A was excavated, allowing its overall ground-plan to be determined with reasonable certainty. Medieval robbing of the building was found to have been extensive and flint wall foundations only survived to any degree in the southern quarter. Much of the outline of the structure was represented solely by robber-trenches, most of which were shallow and had been subsequently truncated by ploughing and/or landscaping (Plate III). From what little evidence survived, it would appear that the complete building belonged essentially to a single phase of construction, and dated no earlier than the late second century ad. Hypocausted wing-room in Building A (Room 4), looking north-east PLATE III KEITH PARFITT 110 The building had been symmetrically laid-out upon a north-east by south-west axis, following the natural fall of the ground, with maximum overall dimensions of 38.50m (ne-sw) by 18m (nw-se). In all, there was evidence for 17 rooms, with one of hall-like proportions at the centre (Room 6). This central hall measured about 5 x 12m and was surrounded by a series of other rooms and corridors. Projecting from the western corner, Room 4 contained the remains of a poorly built, pillared hypocaust, which could have been a later insertion into the original structure (Plate III). Finds of painted wall plaster, abundant roofing tile and some window glass within the associated robber-trenches and demolition deposits indicate that Building A had been a structure of reasonably high status. However, its overall size, the occurrence of only one heated room and a general lack of evidence for the existence of tessellated or mosaic floors, implies that it was not a building in the grandest Roman style. Discussion Although total excavation of the Roman villa complex was beyond the available resources, sufficient work has been undertaken to make it clear that at least two separate, successive flint-built structures existed here. These appear to represent domestic dwellings which had been erected upon an earlier settlement site established during pre-Roman times. The villa buildings presumably lay at the heart of a larger Roman farming estate. No other buildings were located during the present investigations but it may be that not far away once stood a bath-house, barns and other farm out-buildings. It seems quite clear that more detailed and extensive excavations would reveal that the villa buildings formed just one part of a complex, multi-period site spread over a considerable area and generally buried at only a shallow depth. The relatively high water-table implies that significant organic remains may be preserved in waterlogged fills of some of the deeper features. As a Roman villa site, Hull Place is of particular interest because of its low elevation (generally less than 2m above od), being situated just above the marshlands associated with the Lydden Valley. These marshes once formed part of the southern end of the Wantsum Channel, which separated the Isle of Thanet from mainland Kent during the Roman period (Fig. 1). Based upon a consideration of the local coastal morphology and their documented early existence in the medieval period, Dr Steven Fuller has recently conjectured that the reclaimed marshes in the southern part of the Lydden Valley may have been first drained during Roman times (LVRG 2006, 35-6). Clearly, a local context for any Romano-British use of the marsh is needed and the existence of a villa complex at Hull Place close-by, EXCAVATIONS AT HULL PLACE ROMAN VILLA, SHOLDEN, 2005-7 111 potentially represents an important detail. A key research objective for the present investigation, therefore, was to attempt some assessment as to whether the occupants of any Roman site at Hull Place might have been of sufficient wealth and status to have undertaken a major land-reclamation project on salt-marshes in the adjacent Lydden Valley. On the strength of the excavated stone buildings,��������������������������,whichbothshowevidence which both show evidence of some sophistication,���������������������������������������������������,therenowseemslittledoubtthattheowner(s) of there now seems little doubt that the owner(the Hull Place villa complex, occupying a site established well before the Roman Conquest, had considerable means and quite probably could have afforded to undertake marshland drainage works, if so inclined. However, this is not to say that such was the case and the villa estate might equally well have made very profitable use of the adjacent marshland in its un-drained, natural salt-marsh form. Reclamation of marshlands certainly did occur during the Romano-British period, notably in the East Anglian Fens, the Somerset Levels and along the coast of South Wales. ��������������������������������������Asyet,thereisnodefiniteevidenceAs yet, there is no definite evidence for any Roman activity in the marshes of the Lydden valley but perhaps the arrangements in the North �������������������������������������������SomersetLevelsprovideaparallel.OntheSomerset Levels provide a parallel���������.Onthe. On the Levels, extensive marshland reclamation seems to have been carried out by local villa estate-owners. Indeed, such work may have been co-ordinated from the villa site at Wemberham, actually situated within the Levels (Rippon 2000, 194). In east Kent, could the Hull Place villa have served as a similar base from which local marshland was drained? Certainly, the villa’s unusual but deliberate positioning at such a low elevation, close to what was probably salt-marsh by Roman times, strongly suggests that some exploitation of these wetlands occurred, within a region that was generally well-populated throughout the Roman period. If drained, these marshes could have provided extensive sheep pastures, just as they did in the medieval period. acknowledgements Thanks are due firstly to Ian Steed, the landowner and his staff, for readily allowing access to the site. Members of the newly formed Lydden Valley Research Group, particularly Dr Stephen Fuller, must be acknowledged for re-kindling interest in this little known Roman site and for their interest and encouragement throughout the course of the project. Of the Dover Group members who undertook the work special thanks are due to Mrs Tina Parfitt and Messrs Geoff Halliwell, Richard Hoskins, David Holman and Howard Jones for their hard work over many weekends. Brian and Carol McNaughton were responsible for undertaking the initial geophysical survey of the site. KEITH PARFITT 112 bibliography Chapman, H.S., 1921, The Story of Dola. Julius Caesar’s Landing Place (London). Fordham, S.J. and Green, R.D., 1973, Soils in Kent II. Sheet TR35 (Deal)(Soil Survey record No. 15). Igglesden, C., 1935, A Saunter through Kent with Pen and Pencil, 30, 60-3. LVRG, 2006, The Geology, Archaeology and History of Lydden Valley and Sandwich Bay (Report by the Lydden Valley Research Group). Parfitt, K., 1980, ‘Roman Remains Discovered at Hull Place Sholden’, Kent Arch. Rev., 60, 247�����-248.-����248.248. Parfitt, K., 1986, ‘The Roman Villa at Sholden, near Deal: some Further Information’, Kent Arch. Rev., 85, 104-105. Parfitt, K., 1992, ‘Observations on the Hull Place, Sholden Foul Sewer Scheme’, Kent Arch. Rev., 108, 174-176. Phillips, A.S., 1964, O.S. Record Card, TR35 SE4. Rippon, S., 2000, ‘The Romano-British Exploitation of Coastal Wetlands: Survey and Excavation on the North Somerset Levels, 1993-7’, Britannia, XXXI, 69-200. Stebbing, W.P.D., 1937, The Invader’s Shore (Deal).