A Roman Occupation Site at Dickson's Corner, Worth

A ROMAN OCCUPATION SITE AT DICKSON' S CORNER, WORTH KElTH PARFITT The coastal marshlands of the Lydden Valley, north of Deal, once formed the southern part of the Wantsum Channel, which separated Thanet from mainland Kent until medieval times; now as drained marshes they are protected from the sea by a natural sand spit, bounded in part by a shingle bank on its seaward side (Fig. 1). Formed by long-shore drift from the south (Robinson and Cloet I 953; Hawkes 1968), the date at which the spit first began to grow out into what would have originally been the open waters of the Wantsum is not certain but it is fairly clear that the sand spit was in existence well before the outer shingle bank started to accumulate. The spit continues its northward growth today, pushing the mouth of the present River Stour ever closer to Than et. Landward of the spit is a broad zone of wind-blown sand before the marshlands proper begin (Shephard-Thorn 1988, 36-7). Throughout the sand-dunes between North Deal and the Sandwich Bay Estate (known locally as the Sandhills or Tenants Hills) there have been occasional discoveries of Roman material over many years (Parfitt 1982). The most extensive remains were discovered and investigated by Dover Archaeological Group (DAG) in 1978 -79 and these form the subject of the present report. During the violent storms of January I 978 part of the Lydden Valley marshes was inundated by the sea. The earthen bank of the inner sea-wall was breached in several places, including a point between Dickson's Corner and Lyddcourt Stile, some 3km north of Deal (Fig. 2). This breach was quickly repaired with material dug from a 170m long ditch excavated along the wall's seaward side. On inspection a member of DAG discovered several Roman pot-sherds, including an almost complete mortarium (Fig 11.4) in the side of this newly cut ditch (Parfitt 1980). Initial archaeological recording work and limited excavation in the area of this discovery was conducted between May and September 1978, prior to the re-filling of the 107 KEITH PARFITT THANET KEY 􀀂 Blown sand m sand & Shlng1e 􀀃 Alluvium • Fort ♦ Bulldin11 ... VIiia ,. Temple 0 Pottery kiln 0 s 10 - Road km Fig. 1 Map of north-east Kent showing location of Dickson's Corner in relation to other Roman sites and local geology emergency ditch. A research project was undertaken the following year, under the direction of the writer and Ben Stocker, involving the excavation of a series of trial-pits to establish something of the nature and extent of the site. 108 A ROMAN OCCUPATION SITE AT DICKSON'S CORNER, WORTH Field 1978 Ditch --► 100 Metres Fig. 2 Overall site plan showing position of Pebble Areas A-J, test-pits and the suggested extent of Roman occupation, with inset location map showing the site in relation to the Sandwich Bay Estate 109 KEITH PARFITT The site is located amongst coastal sand-dunes in the parish of Worth, some 2.6km east of the parish church. (Centred upon NGR TR 362 566). The main area investigated is situated to the south-west of Dickson' s Corner in an unploughed field of rough pasture. The field is triangular in shape, bounded on the south-west and north-west by the inner sea-wall and on the eastern side by a track running between Deal and the Sandwich Bay Estate (Fig. 2, inset). Some excavation was also conducted in the field on the opposite side of the track and on the inland side of the sea-wall (Fig. 2). The gently undulating land surface in this area stands at an elevation of 3-4m above OD, with the present shore-line lying about 400m to the east. The natural sub-soil across the site is wind-blown sand, classified by the Soil Survey of England and Wales as belonging to the Sandown Series (Fordham and Green 1973, 86, 97). It would seem that evidence for a Roman site in this area was first revealed in 1899 when pottery was found by Christopher Solly whilst engaged in building the inner sea-wall at Dickson's Corner (Chapman 1921, 120; Clarke 1964). THE EXCAVATIONS The first phase of work on the site was concerned with an examination of the sides of the 1978 ditch in order to establish the precise context of the initial Roman pottery finds. Subsequently, in 1979, an extensive series of test-pits, generally one metre square, was dug on either side of the sea-wall, in order to establish the extent of the Roman deposits revealed in the ditch (Fig. 2). The presence of livestock, for the most part, prevented more extensive excavations, but it was possible to open three small areas (designated here the South-East, North-East and North-West Area Excavations), in the field immediately to the south-east of the sea-wall in order to investigate promising features. The test-pitting suggested that on the south-eastern side of the sea-wall Roman deposits covered an area measuring about 130m (NW-SE) by S0m (NE-SW). These deposits ran under the sea-wall and continued for about another 130m inland, making the total length of the main axis of the site (i.e. NW by SE) roughly 270m. The Roman occupation area thus appears to cover more than a hectare and has a somewhat linear form, which most probably reflects the topography of the sand spit. The deposits producing Roman material were mainly either dumps of stiff grey-brown sandy clay, or grey-black occupation layers, corn- 110 A ROMAN OCCUPATION SITE AT DICKSON'S CORNER, WORTH posed of carbon and dirty sand, but also included a number of areas of deliberately laid beach pebbles (Fig. 2; Pebble Areas A - J). Contrasting markedly with the clean yellow natural sands, all these deposits were clearly the product of human habitation; many produced pottery and other Roman material. Directly under the turf, the stratified Roman deposits were sealed by a layer of sand, generally grey-brown in colour, and varying in thickness from 0. 10-0. 70m but averaging 0.30-0.40m [contexts DC 2, 10, 17, 19, 24, 39 & 45]. This sealing deposit, probably mainly of wind-blown origin, produced over 1,200 sherds of Roman pottery. derived from the occupation deposits below. together with twentyfour medieval sherds, indicative of casual activity in the area during the eleventh-thirteenth centuries (Fig. 11.8). The underlying Roman deposits reached a maximum thickness of 0.85m in places and everywhere rested on natural sand. The 1978 Ditch (Figs. 2, 3 and 4) [contexts DC 1-15) The emergency ditch excavated in 1978 first led to the discovery of the site and provided a useful long-section through the Roman deposits (Fig. 2). Most of its length was found to be devoid of archaeology, the sides showing a continuous profile of natural sand. At the south-western end, however, a fairly complex series of discoloured sands, dumped clay and pebble deposits were recorded and these produced a significant quantity of Roman pottery and other finds. On the south-eastern side of the ditch, where the mortarium was found initially, the sides of the ditch were cut back to a near-vertical face, and the exposed stratification recorded (Fig. 4). Under the turf. 0.40-0.50m of yellow-brown sand with humus streaks [DC 2) sealed a deposit of stiff grey-brown sandy clay [DC 3]. The upper sand layer yielded more than seventy pot-sherds, mostly first - second century Roman but also including ten medieval pieces, together with small amounts of animal bone, marine shell and twenty-six prehistoric struck flints. The underlying clay [DC 3] contained about thirty-five more sherds of Roman pottery (Assemblage 3), a few pieces of oyster shell, animal and fish bone and a further nineteen struck flints. Also incorporated into this layer, inverted on the surface of the underlying sand [DC 4] had been the substantially complete mortarium. This is of late first-early second century date (Assemblage 2; Fig. 11 .4). The clay deposit [DC 3] appeared to represent some sort of laid floor, and was further examined in the NE Area Excavation, immediately to the south-east (see below; Fig. 3). In the present section it 111 KEITH PARFITT 0 3 4 5 Metres Fig. 3 Plan of the North-East Area Excavation and adjacent 1978 ditch showing extent of clay floor [DC 3] (hatched) was separated from a series of thin, black carbon deposits [DC 6] by 0.20-0.40m of yellow-brown sand [DC 4], which produced part of the skeleton of an ox but no datable finds. A pair of metatarsals from the ox skeleton showed marked splaying of the distal condyles with severe eburnation on the distal face of the medial condyles plus extensive exostoses and remodelling, especially on the dorsal and medial surfaces of the shaft adjacent to the condyles (Reilly I 984 ). Such symptoms describe the presence of osteoarthritis (Baker and Brothwell 1980, 59), a condition which could be caused by the use of this animal for traction (Armour-Chelu and Clutton-Brock 1985, 297). The underlying carbon layers [DC 6] totalled about 0.15m in thickness and rested upon the natural sand [DC 7]. They produced about sixty sherds of pottery datable to the first century AD (Assemblage l ), together with forty small pieces of animal and fish bone, half of which had been burnt. Cutting through the carbon layers [DC 6] but sealed by the sand [DC 4], was a slot or gulley of uncertain purpose; it was rectangul􀁮r in section, 0.21 m wide and 0.09-0. l 2m deep. Aligned north-south, it was traced for a distance of 1.23m from its southern terminal, before being destroyed by the 1978 ditch. The slot was filled with clean yellow sand, devoid of finds. Despite its resemblance to a beam-slot, there was no evidence that it formed part of a structure. 112 NE Section no. I E3□=3::::E::ot::===:I====::!:::====3:====J2 Metres Fig. 4 Section across the south-east face of the 1978 ditch showing clay floor [DC 3) and carbon deposits [DC 6] KEITH P ARFI'IT The layers on the north-western side of the five-metre wide emergency ditch could not be easily correlated with those recorded opposite. The Roman deposits here were covered by 0.10-0.30m of blown sand [DC 10, equivalent to DC 2] which produced about fifty first - second century Roman pot-sherds and a few sherds of medieval date (Assemblage 5), together with a small amount of animal bone and two prehistoric struck flints. The sand sealed an extensive deposit of grey and black sandy clay [DC 11, 12 and 13] about 0.20m thick, which yielded another 109 late first - second century Roman pot-sherds (Assemblage 4 ), a small lead weight (Fig. l 0.6 ), a honestone and a few animal bone fragments. Laid on the clay at one point was a small area of chalk and flint rubble, measuring some 0.63m (SW-NE) by at least 0.50m (SE-NW). Towards the south-western end of the section examined the sandy clay layer [DC 13) sealed a sequence of earlier deposits, mainly grey and brown sand layers, containing much carbon and pockets of small pebbles [DC I 4 and I 5]. These dipped gently down to the south-west, reaching a maximum thickness of 0.60m but failed to produce any datable finds. The North-East Area Excavation (Fig. 3) [contexts DC 2-3] This was excavated on the south-east side of the 1978 ditch in order to further examine the possible floor [DC 3] that had been previously revealed in the ditch section (see above). Investigation showed that this clay deposit measured some 6.1 Orn (NE-SW) by a minimum of 2.00m (NW-SE), the north-western side being removed by the ditch (Fig. 3). Its surface was slightly undulating and lay at 2.75-3.00m above OD. Although not excavated here, where seen in the ditch section, the layer was 0.14-0.25m thick (Fig. 4). A series of small sand-filled holes cutting the clay, that initially had every appearance of being stake-holes, must have been dug by animals on the evidence of their irregular form when excavated. A more certain feature cutting the surface of the clay was a circular pit some 0.55m in diameter and 0.15m deep. This was filled by clean yellowbrown sand and large flints but produced no datable finds (Fig. 3). The purpose of this clay layer remains uncertain. There was no clear evidence that it formed part of a building, although this is certainly possible. However, in the absence of any other obvious structural remains such as walls, post-holes, beam-slots or hearths, it may perhaps be best interpreted as some sort of open-air hard standing, laid over the soft sand. The available pottery dating evidence (see pages 128-9) suggests that the floor was laid in the early-mid second century AD. 114 A ROMAN OCCUPATION SITE AT DICKSON'S CORNER, WORTH The North-West Area Excavation (Figs 5, 6) [contexts DC 24, 45-50; Pebble Area El The NW Area Excavation was centred on a well-defined layer of small flint pebbles set in clay [Pebble Area E, DC 49], located in the south-east side of the 1978 ditch about 25m south-west of the NE Area (Fig. 2). The pebbled area appears to have originally been roughly sub-rectangular in shape, although its northern part had been destroyed by the ditch (Figs 5 and 6). As excavated, it measured 4.50m (E-W) by a minimum of 3.50m (N-S). Its surface was fairly level and lay at about 2.55m above OD. The pebbling was 0.04-0.0Sm thick and contained a few Roman pot-sherds broadly datable to the period c. AD 70-170 (Assemblage 6). Directly over the pebbling were several layers of sandy clay (Fig. 6), totalling 0.30m in thickness [DC 47). These produced twentyeight Roman pot-sherds, dated to the late-first - early-second century (Assemblage 6), and three prehistoric struck flints. The clays were sealed by a deposit of blown sand [DC 45] up to 0.75m thick, which yielded one hundred fragments of first - second century Roman pottery, together with several medieval sherds; also a few pieces of burnt clay, animal bone and eleven more prehistoric struck flints. It seems fairly certain that the clay layers represent successive floor surfaces to which the pebbling had originally provided a base. At one point, however, the pebbles had been slightly burnt before the first clay deposit was laid, implying that the surface of the pebbles actually formed the primary floor. Part of the southern margin of the pebbled area had a very straight edge such as would have been formed if the stones had originally been laid against a wooden plank or beam, now decayed. An oval post-hole [Feature 1 ], measuring 0.24m by 0.35m and 0.20m deep cut the pebbling adjacent to this straight edge and a second [Feature 2) cut the edge of the pebbling near the southwest corner. This was also oval in shape, measuring 0.26m by 0.30m and 0.40m deep. Its filling of grey sand contained seven large flint cobbles, probably packing-stones. Pebble Area E mainly rested on a thin spread of dumped greybrown sandy clay, which covered an area rather more extensive than the pebbling. Presumably a levelling layer, this dump produced no datable finds. It appeared to fill a series of small depressions in the surface of the soft underlying natural sand. These did not seem to represent post-holes and they may be nothing more than disturbances caused by the feet of passing inhabitants or the builders. One hollow [DC 50] yielded thirteen fragments of burnt clay. Immediately to the south-west of the main pebbled area, a black 115 􀀉--·-·-·-. ! ·xxxxxxxxx lxxxx xxxxxx ·xxxxxxxxxx>. / x X X X X X X X X XPosrholeF.2 . x X X X x@x X X X X A IX X X X X X X X X X X X X .xxxxxxxxxxx xxx •.: ii L X X X X X. X X X X X X X X X ><,... • - • ..Jlb,, • Limit or excavarion 1>< X X .X X X' 1 􀀇-􀀈 r·-·! • Trench 4 I I 0 2 Trench 3 --·---􀀉 Section no. 57 Pebble area J 3 4 5Metres Fig. 5 Plan of the North-West Area Excavation and adjacent test-pits showing Pebble Area E and its associated features, carbon deposit [DC 46] and Pebble Area J .......... -..J [22) Clay l{:)1j Clayey sand , .. :.r.:j Flint pebbles V;a Sandy clay I$􀀄• J Chalk pebbles 0 Metres Fig. 6 Section across the North-West Area Excavation showing clay floors overlying Pebble Area E [DC 49) ► ;;o 0 􀀁 􀀂 8 C: 􀀃 0 z en ::j tl'I ► --i t1 () 􀀄 0 :z; 2 en () 0 ;;o z tl'I ?' 􀀂 0 ;;o 􀀃 KEITH PARFITT deposit of clayey sand containing much carbon and burnt red clay occurred [DC 46] (Fig. 5). This fairly certainly represents dumped hearth material. The deposit was up to 0.25m thick and contained over 400 sherds of pottery, including a stamped samian base (Assemblage 7). Much of this pottery was burnt or re-fired and may be the product of salt manufacture (see below). There were also several small pieces of roof-tile, a fish bone, a few oyster shell fragments, nine prehistoric struck flints and a cut fragment of human cranium. The deposit rested on a layer of yellow-grey clayey sand [DC 48] that dipped gently down to the south-west, apparently following the contour of the natural sand below. It produced three Roman pot-sherds (Assemblage 7) and fourteen fragments of fish bone. It seems reasonably certain that Pebble Area E and its associated features represent the remains of a building. This was probably a square or rectangular timber-framed structure, originally measuring some 4.50m by at least 3.50m. The succession of overlying clay floors implies that the structure had been rebuilt several times and since the later floors did not correspond precisely with the shape of the original pebble surface, it may be suggested that the overall dimensions of the structure changed through time. Taken together, the floor sequence and the hearth dump adjacent indicate that the structure was in use for a significant number of years. The pottery dating evidence (page 131) suggests that this was during the second century but it perhaps began in the late first century. Quite possibly, the building was a small dwelling, rebuilt several times on the same site. The re-fired pot-sherds recovered from the hearth dump could be connected with salt production; if this is correct, an industrial function for the adjacent structure might be implied, but small-scale salt production may well have been a craft activity regularly conducted in and around the house. Another area of pebbling, Area J, was located about 1.50m to the east of Pebble Area E (Fig. 5). This was not examined in any detail and its purpose and relationship to Area E remain unclear. Nevertheless, their close proximity suggests that these areas might have been connected in some way, perhaps forming parts of adjacent structures. Area J was also sealed by a deposit of clay that could represent a floor surface (see Table 1). The South-East Area Excavation (Figs 7, 8, and 9) [contexts DC 39 - 44; Pebble Area C] This excavation was located about 45m to the south-east of the 1978 ditch (Fig. 2) and was dug to investigate an area of laid beach pebble 118 A ROMAN OCCUPATION SITE AT D1CKSON'S CORNER, WORTH (a) I I . I i . . I I i' f • • i. . . (b) ·-;.-\ \ \ I ,\, \ \ \ \ \ \ Hearth 2. \ ;, Fig. 7 Plan of the South-East Area Excavation showing (a) the extent of Pebble Area C [DC 40] and (b) the extent of clay floor [DC 43] (hatched) with its associated features, and underlying Hearth 2 [DC 44] 119 KEITH PARFITT found very close to the surface in one of the initial test-pits. Subsequently, this pebbled area (Pebble Area C) was completely excavated and was found to be roughly kidney-shaped (Fig. 7a). Its main axis was north-east by south-west and it measured about 5.00m by 2 .85m. The pebbling was up to 0.25m thick. It was sealed by 0.15- 0.35m of grey-brown sand [DC 39], which supported the modern turf and produced more than 900 sherds of Roman pottery ranging in date from the second- to early third-century (Assemblage 11 ). A small quantity of Roman roof-tile fragments, thirty-five prehistoric struck flints (including a polished Neolithic axe fragment) and seven small lead weights (see page 127) were also recovered. The pebble deposit consisted mostly of local flint beach shingle but also included some chalk pebbles, a number of fragments of quernstone and much broken pottery, all set in brown sand [DC 40]. The well-defined surface had a marked convex camber in all directions, the highest point lying at an elevation of 3.45m OD (Fig. 8). A probable post-hole about 0.30m in diameter and 0.29m deep, packed with four large flint cobbles, was located on the northern edge of the metalled area. A mixed assemblage of almost 600 sherds of Roman pottery, mostly ranging in date from the late first- to the late-second century, was recovered from the pebbling (Assemblage 10), suggesting that it was probably laid around AD 170. Also discovered were a few pieces of Roman tile, a quantity of iron fragments (including part of a possible horse-shoe), a small amount of animal bone, quernstone fragments, three lead weights and twenty-two prehistoric struck flints. The pebbling partially rested upon a sterile dump of brown sandy clay but mainly lay over a 0.10-0.20m thick layer of grey-brown sand [DC 41], which produced over 400 pieces of late-first - secondcentury pottery (Assemblage 9), together with a pair of copper-alloy tweezers (Fig. 10.2), a copper-alloy nail cleaner (Fig. 10.3 ), another small lead weight (Fig. 10.5), some animal bone and thirteen prehistoric struck flints. Extending under both these layers was a continuous deposit of natural clean yellow sand [DC 42], 0.05-0.1 Sm thick. This produced just two early Roman pot-sherds and a piece of animal bone. It seems fairly clear that the upper sand deposit [DC 41] had developed from this underlying natural sand, incorporating much dumped domestic rubbish into its composition. Sealed below natural sand layer [DC 42] was a fairly level deposit of hard brown sandy clay [DC 43], apparently representing a laid floor (Fig. 7b). This seemed to be roughly D-shaped in plan and measured a minimum of 4.70m (NW-SE) by 3.70m (NE-SW). The surface of the clay lay at an elevation of about 3.05m OD. It was 0.07- 0.1 Orn in thickness and consisted of two distinct layers, the upper one 120 NW Section no. 59 Fig. 8 Section across the South-East Area Excavation showing Pebble Area C [DC 40) and underlying clay floor [DC 43] (key on Fig. 6) SE KEITH PARFITT a plain brown sandy clay and the lower one, brown sandy clay with chalk pebbles. In places, the lower layer was visible on the surface, the upper layer being missing and in others it was absent with the upper layer accounting for the full thickness. Thus, it would seem that these were separate elements within a single phase of deposition (Fig. 8). No datable finds were recovered from either layer. On the south-western side, the clay floor was cut by a hearth-pit [Hearth I, see below]. There were also three other features which seemed to be post-holes [Features 2, 4 and 5]. Only one of these [F. 2] was visible on the surface. The other two were filled with brown clay and chalk pebbles, identical to the material used for the main floor and they were not located until the floor was being removed. It seems likely that their associated posts had been removed at an early stage and the resulting holes in-filled with clay. Standing slightly above the general level of the surrounding floor (Fig. 9), the hearth was set in an oval pit, measuring 1.02m (NE-SW) by 0.86m (NW-SE), with steep sides and a flat base. The pit was O. 14m deep and had been carefully filled with small flint pebbles, capped by a layer of clay that had subsequently been heavily burnt. There were no associated finds. The number and arrangement of the post-holes (Fig. 7b) does not provide sufficient evidence to suggest that they formed part of a timber building, to which the clay deposit could have served as a floor but the presence of the hearth is highly suggestive of this. However, the floor may represent an open-air hard-standing, as already suggested for the clay examined in the NE Area Excavation [DC 3]. Either way, the clay presumably provided a firm walking-surface NW m Bumtclay mB]s and . • Flint pebbles O.D.=3.I0m. 25 0 ba E--3 e---d 25 I SE 50cm. I Fig. 9 Section across hearth associated with clay floor [DC43] (Hearth I) 122 A ROMAN OCCUPATION SITE AT DICKSON'S CORNER, WORTH around the hearth. The post-holes perhaps indicate the position of supports for an associated wind-break, or drying racks, rather than walling. The clay floor mostly lay on natural wind-blown sand. On the south-east side, however, it rested on a thin deposit of black ash and carbon (represented by diagonal crosses on plan, Fig. 7b). This produced no finds but was found to be associated with an earlier hearth [Hearth 2, DC 44], which was located near the east corner of the excavation. This hearth was sub-rectangular in shape, measuring 0.60m (NE-SW) by 0.40m (NW-SE). It consisted of burnt brown clay over large flint pebbles. A large sherd from a first-century AD combdecorated, bead rim jar (Assemblage 8; Fig. 11.13) had been incorporated into the pebbles. The hearth itself rested on natural blown sand and clearly represented the earliest feature in the excavated sequence. Other Pebble Areas A total of ten areas of deliberately laid flint beach pebbles was discovered across the site (Fig. 2) [Pebble Areas A-J]. Eight of these were located on the south-eastern side of the sea-wall with two on its north-western side. Just two were fully investigated [Areas C and E, see above] and only very limited examination of the others was undertaken. Table 1 summarises the basic information recorded. It would seem that a range of shapes and sizes are represented and there is no regularity to their distribution or form. Individual areas of pebbling were buried at depths of 0.1 0- l .40m. Pebble Area G, on the inland side of the sea-wall, could be clearly seen on the surface as a low, circular mound in a field of improved pasture. Four of the pebble areas [B, C, E, and I] contained broken Roman pottery within their make-up but only Area C contained more than a few sherds (see page 120). Quernstone fragments had been incorporated into Areas C and D. All but three of the areas [A, F and JJ sealed earlier deposits containing Roman material. The function of the pebbled areas is not immediately obvious. They certainly are not natural and must have been deliberately laid. The three most likely interpretations for them are; floors of buildings; platforms upon which buildings stood; or casual metalled areas (?between buildings). It seems probable that the first interpretation applies to Pebble Area E, which yielded evidence of at least two associated post-holes and a probable wall-timber impression. The pebble layer here seems to have formed the primary floor of a building and was subsequently covered by successive floors of clay. Partially examined Areas D, F and J were also overlain by clay 123 KEITH PARFITT TABLE 1. RECORDED DETAILS OF THE PEBBLE AREAS A-J (see Fig. 2) Depth Shape Length Width Thick- Cxt ness No. A 0.28 ?oval 16.00* 1.00* 0.18- - 0.25 B 0.34 ?oval 8.00* 5.00* 0.10- 51P 0.20 C 0.15- kidney 5.00 2.85 0.10- 401' 0.35 0.25 D 0.70 ?oval 5.00* 3.00* 0.05- 28 0.16 E 0.70 sub-rect 4.50 3.50* 0.04- 49P 0.08 F 1.41 ? 1.00* 1.00* 0.11 - G 0.10 circ. 12.00 12.00 0.10 - H 0.18- ? 3.00* 1.00* 0.05 - 0.45 I 0.50 ? 1.50* 1.50* 0.05 68P J 0.60 ? 1.50* 1.50* 0.08 - (All measurements in metres: depth is below present ground level: Cxt = Context number: * minimum figure: P pottery rec overed) deposits and might have had a similar function. The close proximity of Areas J and E could suggest that they were related in some way. Pebble Area C was distinguished by the large amount of broken pottery and other domestic rubbish that had been incorporated with the pebbles, clearly indicating that it had been constructed on a well-established settlement site. This was in marked contrast to the other pebbled areas, which generally contained only a little or no such domestic debris. The irregular 'kidney' shape (Fig. 7a), along with its cambered surface (Fig. 8) make the interpretation of Area C as the floor to a building somewhat unlikely: it was associated with just a single post-hole. Whether any of the areas served as a hard-standing upon which buildings stood cannot be certain but this seems quite likely. The possibility that other pebbled areas formed localised metalled surfaces between buildings or served as outdoor working areas also seems probable, particularly where the pebbles had been laid directly onto soft sand. The apparently linear form of Area A suggests that it could have formed part of a track-way or path. Utilizing readily available local raw materials, it seems probable that the various 124 A ROMAN OCCUPATION SITE AT DICKSON'$ CORNER. WORTH pebbled areas recorded actually had a variety of different functions. Only large-scale excavations will provide further clues as to their interpretation. THE FINDS The site produced a considerable quantity of Roman finds (Figs 10-12), mostly pottery but also a small collection of metal-work and a number of stone objects mainly quernstone fragments, as well as two hones and a group of prehistoric struck flints. The SE Area Excavation produced a large percentage of all the material recovered. The following notes are by the writer unless otherwise stated. The finds have been deposited at Dover Museum, with a copy of the site archives. Metal and Glass Objects Five copper-alloy objects were recovered (Fig. I 0, Nos. I- 4), all in a moderately poor state of preservation. Only one brooch was found. (I) Complete but distorted wire bracelet with round cross-section. Each terminal has been decorated with seven transverse grooves. Grey-black sand and carbon layer, Trench 8 (Fig. 2). [DC 23] (2) Pair of tweezers with a surviving length of 48mm. The undecorated blades are slightly flared but the in-curved tips have largely broken away. Sand and domestic rubbish layer under Pebble Area C. [DC 41) (3) Nail cleaner with tapered shaft; both ends have corroded away and the surviving length is 35mm. There is a stamped diagonal cross near the top on one side and on the other a longitudinal groove runs down the centre of the lower half. Found with tweezers (2). [DC 41] (4) Corroded brooch of 'Nauheim derivative' type with a high arched bow of circular cross-section. Spring of four coils; catch plate missing. First century AD. Grey sand layer, Trench 7 (Fig. 2). [DC 22] (5) (Not illustrated) Stud with a flat head and bent shaft. The shaft is 13mm long and tapered, with a square cross-section. The corroded head was probably originally circular, at least 19mm across. Sand over Pebble Area C. [DC 39] One hundred and six pieces of iron (not illustrated) were recovered from the site, all in a poor condition. Just over three-quarters of the finds came from the SE Area Excavation, particularly the make-up of Pebble Area C, which contained 33 items. Most of the objects appear to be nails of small to medium size, with rounded heads and square sectioned shafts. There are also several fragments from fittings, all of indeterminate form. Part of a possible horse-shoe came from Pebble Area C but a number of small, shapeless masses of corrosion are beyond all identification. No black-smithing debris, such as slag, came from the site, perhaps implying that all the iron items employed here were brought in ready made. 125 KEITH PARFITT i, 2 5 4 - Pti.ftn Fig. 10 Small finds. Scale, I: I. No. 5, weight, 23.Sgm; No. 6, weight, 38.7gm 126 A ROMAN OCCUPATION SITE AT DICKSON'$ CORNER, WORTH The site produced thirteen pieces of lead of which eleven came from the SE Area Excavation. One of the test-pits produced a small circular lump about 55mm in diameter and up to 5mm thick, which appears to have been folded over double and was perhaps intended for re-melting, but all the remaining pieces are of a similar form. These objects were clearly deliberately produced and consist of short strips bent round length-wise to produce a cylinder with a penanular or U-shaped cross-section (Fig. I 0, Nos. 5 & 6). They range in length from 36-55mm, although most are between 43-48mm. The weights of complete examples vary from I6.6-39.7g. Lead objects of this type have been previously identified as small weights probably used on fishing nets. In this context it is particularly interesting to note the edges of one example (Fig. I 0.5) which bears the impressions of where it has been crimped onto some other thin object that had small, regularly spaced protrusions, set at intervals of around 5mm. Perhaps these protrusions were the individual cords along the edge of a net. Comparable finds come from the late Roman fort at Portchester, Hampshire (Webster 1975, fig. 123, no. 167) and the native settlement at Bishopstone, Sussex (Bell 1977, fig. 84, nos. 22 and 23). Although it seems most probable that these are weights from fishing nets. it is possible that such objects could be derived from nets associated with catching birds, particularly wild fowl. A small, thin wall fragment from a vessel of clear blue-green glass came from a black sandy clay layer in Trench 21 (Fig. 2). [DC 30) Pottery (Figs 11 and 12) by Malcolm Lyne The site produced six Gallo-Belgic, 108 samian and 3,995 other sherds of late first to early third century coarse pottery, together with 41 pieces of medieval ware. The samian ware was sent to a specialist for identification but was mislaid and must now be presumed lost. It is recorded, however, that this material included a stamped base [DC 46] and a decorated fragment from a Dr. 29 bowl (c. AD 43-85, unstratified). The table below summarises the distribution of sherds across the site:- TABLE 2. DISTRIBUTION OF POTTERY FROM THE DICKSON' S CORNER SITE (excluding samian ware) Location No. sherds % Unstratified 233 6 1978 Ditch + NE Area Exe. 341 8 NW Area Excavation 565 14 SE Area Excavation 2036 51 Other test-pits 867 21 Total 4042 100 127 KEITH PARFITT The Ga!lo-Belgic and other wares were initially looked at by Nigel Macpherson- Grant soon after the excavation and were re-examined by the author 20 years later. All assemblages were quantified by numbers of sherds and their weights per fabric: these fabrics were classified with the aid of a x8 lens with built-in metric scale for identifying the nature, shape and frequency of inclusions. A x30 magnification pocket microscope with built-in light source was additionally used on the finer fabrics. Two of the pottery assemblages, from DC 39 and DC 40, are large enough for quantification by Estimated Vessel Equivalents (EVEs) based on rim sherds (Orton 1975). All quantifications must be regarded as less than completely satisfactory because of the loss of the samian. Nevertheless, at less than 3% of the total pottery from the site, the loss of this material can only have a marginal effect on the calculated percentages of other fabrics: it does, however, mean that our understanding of the pattern of pottery supply to the site is incomplete. The various assemblages are described below. NE Area Excavation and south-east side of the 1978 ditch This area produced evidence for two successive phases of occupation:- Assemblage 1 From carbon layer DC 6 on the surface of the natural sand This context yielded 64 sherds of late Iron Age to Claudian character, including the following pieces:- Fig. 11.1 Small carinated cup of unusual form in Gal\o-Belgic Terra-Nigra. Exact parallels have proved elusive but a larger version of the form in native coarse fabrics is present in Augustan assemblages from Henin-Beaumont in the Pas-de-Calais and elsewhere in that region (Geoffroy et al. 1998, pls 4-17). Ext. rim diameter 100mm. Fig. 11.2 Cordoned beaker in tournette-finished orange-brown Fabric B2 with grey patches. Paralleled at Canterbury in late Iron Age levels (Blockley et al. 1995, fig. 4-81 ). Ext. rim diameter 80mm. Fig. 11.3 Wheel-turned thin-walled jar of Monaghan ( 1987) Type 4J I in black-cored buff variant of Upchurch Fabric R 16 with profuse silt-sized quartz, occasional mica platelets, sparse 0.50mm multi-coloured quartz and a single 5.00mm calcined flint inclusion. Ext. rim diameter 140mm. Black coring is characteristic of the earliest Claudian Upchurch products. Date, c. AD 43-55. Assemblage 2 From the upper surface of yellow-brown sand layer DC 4, sealing DC 6 The greater part of the following vessel was laid inverted on the surface of the sand:- Fig. 11.4 Mortarium in light-grey Canterbury kilns fabric fired brown-buff. Very worn to the extent of having a hole in the bottom c. 750mm in diameter. The section of rim containing the spout is missing. Date, c. AD 75-125. Assemblage 3 From the grey-brown sandy clay layer DC 3, above DC 4 The 37 sherds of pottery from this layer are too few in number for any kind of meaningful quantification but include Upchurch (R 16) ware rim fragments 128 A ROMAN OCCUPATION SITE AT DICKSON'S CORNER, WORTH C ) I ) \ 3 4 ' 􀀁 I 5 7 􀀄- ) \ ·,;{ 􀀁 ' 8 0 10 􀀂 cm. -􀀃< 10 l \ \ 12 ,- x::--,16 13 Fig. 11 Coarse pottery from Dickson' s Corner 129 KEITH PARFITT from Monaghan's (1987) beaker forms 3A2.2 (c. AD 43-70) and 3B.I (c. AD 50-100) and a dish of form 7 A2.4 (c. AD 43-140). Three latticed body-sherds from a BB2 'pie-dish' do, however, indicate that this otherwise mid-late first-century occupation assemblage continued to be added to after c. AD 120. (The small 76 sherd assemblage from the upper wind-blown sand above this layer [DC 2) is of very similar character but also includes two post AD I 00 (probably post AD 130) sherds from a Cologne beaker, together with ten medieval pieces). North-west side of 1978 Ditch Assemblage 4 From the sandy clay layers DC 11 and 12 These layers produced a total of I 09 sherds of Roman pottery with a similar date range to that of DC 3 on the other side of the ditch (Assemblage 3). Once again the assemblage is too small for meaningful quantification but includes a lid-seated jar rim sherd in rough Canterbury greyware (Pollard 1988, fig. 28.68, c. AD 70-175), a pulley-rim flagon sherd from the same source (ibid., fig. 28.77, c. AD 70-150) and a reeded-rim bowl fragment in Canterbury orange-buff ware (c. AD 70-175). The following pieces are also present:Fig. 11.5 Lid in beige sand-free Pompeian Red fabric PRW3 from Central Gaul (Peacock 1977) with internal reddish-brown colour-coat and golden mica-dusting. The lid has patchy blackening from use. Ext. rim diameter 220mm. Date, c. AD 55-140, particularly common during the period c. AD 120-140 in London (Davies et al. 1994, 134). DC 11. Fig. 11.6 Platter of Monaghan ( 1987) Class 7 A2 in grey Upchurch ware. Ext. rim diameter 150mm. Date, c. AD 43-140. DC 12. The assemblage is dominated by 'Belgic' grog-tempered and Transitional 'Belgic' grog-tempered/Native Coarse Ware jar sherds (62% by sherd count), although there are no drawable profiles. Assemblage 5 From the upper blown sand layer DC I 0, sealing DC I I and I 2 The 55 sherds from this deposit comprise 49 Roman fragments of similar date range to the assemblage from DC 11 & 12 and six medieval sherds. The following pieces are of particular interest:- Fig. 11. 7 Everted rim jar with rim top flattening and a comb-stabbed neck cordon, in patchy black/buff Transitional 'Belgic' grog-tempered/Native Coarse Ware. Ext. rim diameter 155mm. The type is common at Richborough and there dated to the late-first century AD (Bushe-Fox I 926, pl. XXIV, 48). Related comb-stabbed forms have been more recently dated by Pollard to the period c. AD 70-120 ( 1988, fig. 15.39, 40). Fig. 11.8 Rim sherd from cooking-pot in Early Medieval Sandy ware. Date, c. AD 1050-1150. North-West Area Excavation, Pebble Area E Assemblage 6 From Pebble Area E [DC 49] and clay floor layers above [DC 47) The pebble floor itself [DC 49) produced only seven sherds, comprising a tiny chip from an Upchurch beaker and six closed form sherds in Transitional 'Belgic' grog-tempered/Native Coarse Ware. None of these sherds is closely 130 A ROMAN OCCUPATION SITE AT DICKSON'$ CORNER, WORTH datable and all that can be said with any certainty is that the floor was laid at some time during the period c. AD 70-170. The clay floor layer above [DC 47] yielded a further 28 sherds of late-first to early second-century pottery, including the following fragment:- Fig. 11.9 Necked and cordoned jar in rough Canterbury greyware. Ext. rim diameter 1 60mm. This is a long-lived type and it is as yet not possible to date it more closely than c. AD 70-175. A rim fragment from a reeded-rim Canterbury greyware bowl and the carination from an Upchurch ware biconical vessel are also present in this assemblage. Assemblage 7 From clayey sand layer DC 48 and dumped hearth material DC 46 above, SW of Pebble Area E The clayey-sand layer immediately above the natural [DC 48] produced a mere three closed form sherds in Transitional 'Belgic' Grog-tempered/ Native Coarse Ware. In themselves, these pieces are not closely datable but they are oxidised and crackled from over-firing. The 419 sherds from the dump above (DC 46) make-up a large enough assemblage for quantification by number of sherds per fabric:- TABLE 3. DETAILS OF POTTERY FABRICS FROM DC 46 (Assemblage 7) Fabric No. of sherds % B2 Oxidised 1 5 3.6 B2 Black 44 10.5 B2/Rl Oxidised 243 58.0 Rl I 0.2 R5 10 2 .4 R6 35 8.4* Rl4 10 2.4** Rl 6 37 8.8 R73 Grey 11 2.6 R73 Oxidised 7 1.7 Misc 5 1.2 Rll3 1 0.2 Total 419 100 (* may include re-fired R5: ** includes part re-fired sherds) The pottery fabrics have been classified according to the system developed by Canterbury Archaeological Trust for east Kent. The following codes have been used in the tabulated quantifications:- B2 'Belgic' Coarse Grog-Tempered B8 'Belgic' Fine-Sanded Rl Native Coarse Ware 131 KEITH PARFITT R5 Canterbury Coarse Grey Sandy Ware (Flavian/Antonine) R6 Canterbury Coarse Oxidised Sandy Ware (Flavian/Antonine) Rl4 Cliffe BB2 Rl6 Fine Grey Upchurch Ware R17 Fine Oxidised Upchurch Ware R33 Colchester Colour-Coated Ware R50 Baetican Dressel 20 amphorae R73 Thameside Grey Ware RI 13 Gauloise Amphorae LR2.2 'Scorched' Coarse-sanded Thameside Grey Ware LR16 German Marbled Ware One of the most distinctive characteristics of this assemblage is the high percentage of re-oxidised sherds (73%). Although some of the jar sherds in Transitional 'Belgic' Grog-tempered/Native Coarse Ware are crackled like those from DC 48 and have the appearance of wasters, sherds in Canterbury Greyware and BB2, from sources some distance away, are also oxidised. It is unlikely that what we have here is merely due to cooking over open fires, as such activity would be unlikely to heat vessels to the point where they became bloated or oxidised all over. It seems more likely that some kind of industrial process took place nearby and that this is the waste material. Re-fired vessels include bowls, jars and flagons and there are fragments of fired clay ?furnace or oven-lining. Vessel fragments include the following piece:- Fig. 11. I O Base from re-fired jar in very-fine-sanded ?Thameside greyware with hole drilled through off-centre after firing to convert into a strainer. All of this could suggest small-scale brine boiling in pots, straining wet salt and drying it in open bowls and dishes. The presence of a number of unaltered sherds, including finewares, indicates that domestic activity was also taking place close-by. The pottery forms include furrowed and combed body-sherds from 'Belgic' Grog-tempered ware jars (late Iron Age to c. AD I 00), rim sherds from at least seven Transitional 'Belgic' Grog-tempered/Native Coarse Ware cookingpots (c. AD 70-170), a platter of Monaghan's ( 1987) Form 5B2.1 (c. AD 90- 130) and several re-fired 'pie-dishes'. The following pieces are of particular interest:- Fig. 11. 11 Dr. 37 bowl copy of Monaghan ( 1987) Class 4H2 in grey Upchurch ware with compass-scribed 'London ware' decoration. Ext. rim diameter I 80mm, c. AD 70-130 (noted in Marsh 1978, 126). Another fragment of this vessel was present in the wind-blown sand DC 45, over DC 46. Fig. 11. 12 Small jar in rough grey Canterbury fabric with horizontal shoulder groove. Ext. rim diameter 120mm. A date c. AD 70-130+ is indicated for this assemblage. The sherds from the wind-blown sand DC 45 above are similar to those from the equivalent deposit in the NE Area [DC 2] in including an eleventh-twelfth century medieval element. 132 A ROMAN OCCUPATION SITE AT DICKSON'S CORNER, WORTH South-E as t Area Excavation Assemblage 8 From Hearth 2 on natural, DC 44 This hearth yielded two large fresh sherds from the following jar:- Fig. 11.13 Bead-rim jar with combed decoration, in grey, coarse 'Belgic' grog-tempered ware. Ext. rim diameter 190mm. This is similar, both in form and decoration to Canterbury Form 13 (Pollard 1995, fig. 269.15). Date, c. AD I - 80+, probably Pre-Flavian. Assemblage 9 From grey-brown sand DC 41, below Pebble Area C Sand and clay layers DC 42 and 43 above Hearth 2 lacked pottery but DC 41 above them yielded 476 sherds. This assemblage was quantified by numbers of sherds per fabric:- TABLE 4. DETAILS OF POTTERY FABR1CS FROM DC 41 (Assemblage 9) Fabric* No. of sherds % B2 6 1.2 B2/R 1 Oxi dised 5 7 12.0 B2/Rl Black 166 34.9 B8 Black 5 1.1 B8 Oxidised 2 0.4 Rl 22 4.6 R5 15 3.2 R6 4 0.8 Rl4 27 5. 7 Rl6 136 28.6 Rl 7 6 1.2 R50 5 1.1 R71 6 1.2 R73 3 0.6 LR2.2 1 0.2 Misc 15 3.2 Total 4 76 100 * See codes listed under Table 3. Quantities of re-fired pottery are far less than in DC 46 ( 13%) and almost entirely restricted to grog-tempered wares: the sherds may not be re-fired at all but merely from patchy clamp-firing in pottery manufacture. This factor coupled with a considerably higher percentage of Upchurch finewares suggests that this is a straight-forward domestic assemblage. The material includes jars in 'Belgic' Grog-tempered ware, Transitional 'Belgic' Grog-tempered/Native Coarse Ware and knife-trimmed Native Coarse 133 KEITH PARFITT Ware. indicating a date range from the late-first century to after AD I 70. Other forms are a Monaghan ( 1987) Type 5D2.2 bowl in grey Thameside fabric (c. AD 120-180), a flask of Type 3A2.2 in grey Upchurch ware (c. AD 43-70), a bead- rim beaker of Class 3E6 in similar fabric (c. AD 40/50 - I 00) and the following:- Fig. 11.14 Everted rim jar in brown-black 'Belgic' Grog-tempered ware with stabbed decoration on the shoulder. Ext. rim diameter 200mm. Similar to an example dated c. AD 50-80 from Richborough (Bushe-Fox I 949. Pl. LXXXVIII, 391 ), Fig. I 1.15 Bead-rim jar in similar fabric and with similar stabbed decoration. Ext. rim diameter 160mm. Fig. 11.16 Storage-jar with rolled over rim and latticed shoulder decoration, in patchy orange-brown/brown-black Transitional 'Belgic' Grog-tempered/ Native Coarse Ware fabric. Ext. rim diameter 280mm. Similar to an example from the Marlowe Theatre site in Canterbury (Pollard 1995, fig. 304, F14J). Date, c. AD 130-170. Fig. 11.17 Rim from lid-seated neckless jar in rough grey Canterbury fabric. Ext. rim diameter 160mm. Similar to an example from Canterbury (Wilson 1983, fig. 87.172B). Date, c. AD 120-150. Fig. I I. I 8 Flanged dish of Monaghan (1987) Class 7 A2 in oxidised Upchurch ware. Date, c. AD 43-140. Fig. I 1. l 9 Upper part of biconical vessel of Monaghan ( 1987) Type 201.6 in grey Upchurch ware. Ext. rim diameter 160mm. Date, c. AD 70- I 00. Fig. 12.20 Flanged dish with acute lattice decoration of Monaghan ( 1987) Type 5D4.2 in B82 fabric. Ext. rim diameter 200mm. Date, c. AD I J0/120- 160/180. Assemblage JO From Pebble Area C [DC 40]. above DC 41 This pebble layer produced a 598-sherd assemblage, the second largest from the site. It is large enough to be quantified by Estimated Vessel Equivalents (EVEs) based on rim sherds (see Table 5). The breakdown of this assemblage shows a further decline in the significance of grog-tempered wares and a great increase in that of fine Upchurch greywares. There are no products from the Canterbury kilns other than a few flagon and other closed form bodysherds, which may be residual. Forms include biconical vessels of Monaghan's (1987) Classes 20 I (c. AD 60-130) and 202 (c. AD 50-100) in grey Upchurch ware and the following:- Fig. 12.21 Necked jar in handmade brown-grey Native Coarse Ware. Ext. rim diameter 240mm. One of three similar examples. Fig. 12.22 Carinated bowl in similar fabric fired grey-black. Ext. rim diameter 220mm. Paralleled at Canterbury Lane (Wilson 1983, fig. 110-474). Date, c. AD 160-190. Fig. 12.23 Deep pie-dish of Monaghan (1987) Type 5D1.6 in black BB2 fabric but without decoration. Ext. rim diameter 180mm. Date, c. AD I JO/ l 20-150/180. Fig. I 2.24 Plain 'pie-dish' of Monaghan ( I 987) Type SC4.2 in black BB2 fabric. Ext. rim diameter 200mm. One of two. Date, c. AD 150/170-250. 134 A ROMAN OCCUPATION SITE AT DICKSON'S CORNER. WORTH TABLE 5. QUANTIFICATION OF POTTERY from (above) Pebble Area C [DC 40, Assemblage 10} and (below) [DC 39, Assemblage 11 by Estimated Vessel Equivalents (EVE) Fabric Jars Bowls Dishes Beakers Store- Others Total % EVE EVE EVE EVE jars EVE EVE EYE B2/Rl l.23 0.19 - - - - 1.42 22.8 RI 0.70 - - - - - 0.70 11.3 Rl4 0.08 0.85 - - - - 0.93 15.0 Rl6 - 0.14 - 1.40 Rask 1.00 . - Biconical 0.43 2.97 47.7 Misc - - - - - 0.20 0.20 3.2 Total 2.01 1.18 - l.40 - 1.63 6.22 - % 32.3 19.0 - 22.5 - 26.2 - - 82/Rl /.82 - - - - 1.82 22.5 Rf /.60 - . - - . l.60 19.8 R6 0.09 . . . . 0.09 1.1 R/4 0.10 1.22 . . . 1.32 16.3 R16 0.84 . 0.86 0.53 2.23 27.4 R33 - 0.05 . 0.05 0.6 R73 0.67 . 0.67 8.3 LR16 Flagon 0.32 0.32 4.0 Total 5.12 1.22 0.9/ . 0.85 8./0 % 63.2 15.l . 11.2 . /0.5 . . Fig. 12.25 Small everted-rim beaker of Monaghan ( I 987) Class 2H I in grey Upchurch ware. Ext. rim diameter 90mm. Date, c. AD 80/90-120/130. Fig. 12.26 Rim of poppy-head beaker of Monaghan ( 1987) ?Class 2A4. Ext. rim diameter 90mm. Date, c. AD 130-170. Fig. 12.27 Upper part of bottle of Monaghan (1987) Class I 84 in grey Upchurch ware. Ext. rim diameter 90mm. Date, c. AD 70-110. It is noteworthy that the Upchurch finewares tend to date to the late-first and early-second centuries, whereas the coarsewares are mainly of late-second to early-third-century date. This suggests that the inhabitants of the site were of fairly low status and at the time of the laying of the floor around AD 170 were still using up to I 00 year old Upchurch 'table' ware alongside more recent, shorter-lived cooking-vessels. Assemblage / I From the wind blown sand DC 39, over Pebble Area C The 958 sherds from this context constitute the largest single assemblage from the entire site and are mostly derived from the occupation associated with Pebble Area C, DC 40. The assemblage was large enough for quantification by EVEs (see Table 5). 135 KEITH PARFITT 20 0 10 an. 21 1 l ) ' I 22 􀀁 23 D 1 \ ) 2F Jlr 25 27 ) 􀀄􀀅--\ \ 28 29 7 ' I (; \ 30 31 32 33 m 34 Fig. 12 Coarse pottery from Dickson's Corner This occupation assemblage is characterised by an increase in the significance of Native Coarse Ware, a decline in that of fine grey Upchurch ware and the appearance of significant amounts of grey Thameside wares. It can 􀁩e broadly dated to the period c. AD 170-250, although it is unlikely that there 1s anything later than the first two decades of the third century. The form breakdown differs markedly from that of the material built into the floor beneath [DC 40], in that cooking-pots make up nearly two-thirds of all of the pottery. This is the normal situation on low status rural sites and contrasts with the domination of the floor assemblage by fineware beakers and other forms. None of the pots show any significant sign of re-firing and may be regarded as a 136 A ROMAN OCCUPATION SITE AT DICKSON'S CORNER, WORTH normal domestic assemblage. There are a large number of bodysherds from a Dressel 20 olive oil amphora but these may be from an old vessel used for storage rather than implying the use of olive oil in cooking by the inhabitants of the site. The vessels include the cornice-rim from a Colchester Colour-Coat beaker (c. AD 130-200), three examples of 882 pie-dish form 5C3. l (c. AD 150-250) and the following:- Fig. 12.28 Complete rim circumference and greater part of body of everted rim jar in black Native Coarse Ware with polish over shoulder and rim. Ext. rim diameter 180mm. There are fragments from at least ten other similar jars in both Transitional and Native Coarse Ware. Fig. 12.29 'Pie dish' of Monaghan (1987) Type 5C4.4 in black 882 with brown margins. Ext. rim diameter 180mm. Date, c. AD 170/190-210/230. Fig. 12.30 Another example with chamfered base in similar fabric. Ext. rim diameter 200mm. Fig. I 2.31 Beaker of Monaghan ( 1987) Class 2B2 in grey Upchurch fineware. Ext. rim diameter 110mm. Date, c. AD 40/50-90. Fig. 12.32 Upper part of a biconical vessel in similar fabric. Ext. rim diameter 180mm. Similar to example from St Margaret's Street baths, Canterbury (Wilson 1995, fig. 302.F94). Late second century. Fig. 12.33 Rim from jar of Monaghan ( 1987) Type 3H7 .7 in grey Thameside fabric. Ext. rim diameter 140mm. Date, c. AD I 80-250. Fig. 12.34 Pulley-neck flagon of early-third-century type in sand-free pinkishbuff fabric with sparse brown up to 2.00mm argillaceous inclusions and traces of brown slip. Ext. rim diameter 45mm. This fabric is very similar to that of the later German Marbled Ware and may be from the same source. The breakdown of the early second-century pottery assemblage from DC 46 indicates that the bulk of the coarse pottery (72%) came from 'Belgic' and Transitional 'Belgic' /Native Coarse Ware production sites in east Kent. One such production site may have been in the Whitstable area. Small quantities of Romanised sandy grey and oxidised flagons. bowls and other forms ( 11 %) were supplied by the Canterbury kilns and 8B2 pie dishes, Thameside greywares and Upchurch finewares (15%) from coastal production sites in and near the Medway marshes. The late second-century saw the virtual disappearance of Canterbury products from the site and an increase in the B82, Thameside greyware and Upchurch fineware share of the assemblage from DC 39 to 52%. These Medway estuary production sites now provided most of the open forms and fine wares and some of the cooking-pots supplied to the site. This increase in the supply of wheel-turned Romanised wares from the Medway sources seems to have been at the expense of the native industries. Their share had declined from 72% to 42% and was now almost entirely restricted to cooking-pots. It appears that nearly all of the pottery supplied to the site throughout its occupation came from more westerly coastal sources along the Thames estuary, or from Canterbury on the River Stour and may have been delivered by waterborne traffic rather than overland. 137 KEITH PARFITT Chaff Tempered Ware Nine fragments from small, crude ceramic vessels of the type first described by Macpherson-Grant (1980) were recovered from the site. These were scattered singly throughout the deposits and five were unstratified; there are two rim sherds but the overall shape of vessels from which these come remains unclear. The date-range of this ceramic-type seems to be limited to the first century AD. and it has been suggested that the vessels represent salt containers. Given the coastal location of the present site, with its potential for salt production, the general lack of such ware could perhaps be significant. Chronological factors, however, might be equally important and it may be that the bulk of the excavated deposits at Dickson 's Corner are simply later than the production period of vessels in this fabric. Tile and Burnt Clay The excavations produced a total of fifty-nine, mostly small Roman tile fragments, representing a minimum of eleven complete tiles in hard, red fabrics. With the exception of two pieces of imbrex, all this tile appears to be of tegula type. Just over half the fragments recovered came from the sand deposit over Pebble Area C [DC 39]. Pebble Area C itself contained a further four pieces. The NW Area Excavation produced fifteen small fragments, all from the carbon dump deposit [DC 46] south-west of Pebble Area E. The other tile fragments were recovered from various deposits in out-lying test-pits. From the small quantities of tile found and its fairly scattered distribution it seems unlikely that any structures with tiled roofs existed on the site. More probably, the small amounts of such material had been brought to the site for other purposes, such as the construction of hearth bases, etc. Two fragments of medieval peg-rile were recovered from the upper sand deposit of the NW Area Excavation [DC 45]. The excavated deposits produced a total fifty-four small pieces of burnt clay. This material was scattered across the site with no significant concentrations. Most fragments are orange-red in colour, often with traces of burnt-out organic temper, although the filling of a hollow under Pebble Area E [DC 50] produced thirteen pieces in a slightly sandy brown-black fabric. None of the fragments recovered can be positively identified as being either burnt daub or briquetage from salt production. Hones i) Rectangular hone of moderately fine grained light grey-brown, sandstone. One end broken; surviving length, 100mm. Edges rounded by use except for a strip about 20 mm. wide at the unbroken end. From grey sandy clay layer, 1978 Ditch. [DC 11] ii) Broken fragment from a rectangular hone of grey-brown sandstone, slightly more coarse than (i). Surfaces stained brown by iron; surviving length, 51 mm. Grey clayey sand layer under Pebble Area D. [DC 29] Quernstones (not illustrated) by Howard Jones The site produced a total of 65 fragments of stone, certainly or probably 138 A ROMAN OCCUPATION SITE AT DICKSON'S CORNER, WORTH derived from rotary querns. The bulk of the pieces came from the SE Area Excavation: the metalling of Pebble Area C [DC 40] contained forty-one of the fragments and a further eleven were recovered from the overlying sand [DC 39]. The s and and domestic rubbish layer under Pebble Area C [DC 41) produced another four pieces. Elsewhere, a single fragment came from the make-up of Pebble Area D and three from Pebble Area I. Another came from the sand above Pebble Area E [DC 45) and three more pieces were recovered from the sides of the 1978 ditch. The fragments may be grouped into three basic rock types:- !) Brown sandstone, of differing degrees of coarseness, probably representing Millstone Grit from the Pennine region (37 fragments) 2) Greensand from the local Folkestone Beds (7 fragments) 3) Basalt lavas tone, probably from the Mayen district of Germany (21 small fragments) At least twelve separate quemstones are likely to be represented. The surfaces of two pieces had subsequently been partially vitrified by intense heating. The relative paucity of querns made from local greensand is of interest, as work on the foreshore at East Wear Bay, Folkestone, about 23km to the south-west of Dickson's Corner, has demonstrated that quernstones were being manufactured there throughout the late Iron Age and earlier Roman periods (Keller 1989). However, the predominance of Millstone Grit at Dickson's Corner is largely explained by the presence of at least twenty-nine fragments from a single quernstone of this rock-type, recovered from Pebble Area C. These formed about half of a well-worn and unusually large stone, some 780mm in diameter. From the size, this could be more accurately described as a millstone but its thickness (55mm max.) and the presence of a handle hole suggests that this is not the case. Two small fragments, also apparently derived from this stone, were recovered from the sand deposit underlying Pebble Area C [DC 41]. The other quern fragments all seem to be from smaller domestic querns used for the grinding of grain and would appear to be fairly typical of those found on Romano-British settlements throughout east Kent. Struck Flints (not illustrated) by Geoff Halliwell The excavations produced a total of 181 prehistoric struck flints. The SE Area Excavation yielded 70 of these, whilst the NW Area produced a further 23 and the 1978 ditch, 47. Fourteen more were unstratified and the remainder came from out-lying test-pits. Of the stratified material, 88 pieces were found scattered throughout the Roman levels themselves and the remaining 79 were recovered from the sealing upper sand layer. None was contained within the natural sand deposits below the Roman layers. The raw material used includes significant proportions of Bullhead flint, nodules with a thick white cortex and others with a thin grey cortex (? river gravel). Only a few pieces of beach pebble occur. The flints include a number of worked pieces, most notably part of the cutting end of a finely polished axe, found in the sand over Pebble Area C [DC 39]. There are also three scrapers (two end-scrapers and one side- 139 KEITH PARFITT scraper), three piercers, one fabricator/smoother, two serrated flakes (including a 'fish de-scaler', see below) and five utilised flakes. There is one hammerstone made from a naturally perforated pebble, which may have been hafted. No cores were discovered but one core-rejuvenation flake was noted. One calcined flint (pot-boiler) is present. The fifteen worked flakes and implements recovered represent a very limited 'tool-kit', consisting mainly of roughly fashioned 'tools of the moment' with haphazard working and utilised waste flakes. A possible preponderance of piercers for making small holes may be represented but the general absence of knives, consistent scrapers and arrowheads of any kind is noteworthy. The only readily datable implement within the assemblage is the polished Neolithic axe fragment. A large serrated flake is of particular interest. It consists of a secondary flake measuring 93mm by 37mm. Down one side a saw-like edge with seventeen teeth has been very deliberately produced, in a manner somewhat reminiscent of a fish de-scaler on a modern pen-knife. (The number of teeth per inch, 7 or 8, is exactly the same as a present-day Swiss Army knife!). The implement appears to be unique for this part of Kent, although need not be seen as anything more than another quickly made, 'one-off' tool. The significance of the prehistoric flints from the site is difficult to assess; taken at face value, as they would be on most local sites, it could be inferred that limited prehistoric occupation occurred in the area of the site prior to the Roman settlement. In the present location, however, such an interpretation has important implications, for it suggests that the sand spit upon which the site stands had formed long before the Roman period. Before accepting this, other possible sources for the flint-work must be considered. A submerged prehistoric land surface has been recorded under the marshland immediately to the west of the site (Halliwell 198 I; Halliwell and Parfitt I 985) and it is possible that prehistoric material was brought from there to Dickson's Corner in the Roman period. The most obvious way this could have occurred was if the flints had been incorporated in the marshland clays that were clearly being brought to the site for use as floors [DC 3, 43, etc]. However, only 35 (40%) of the 88 flints recovered from the Roman layers were contained within clay deposits. Moreover, the general appearance of the flintwork is not like the material found in the adjacent marshlands. A general decline in the quality of flint working is now widely recognised in the later prehistoric period and the standard of workmanship represented by the present material is suggestive of a poor, late industry. Locally, the material can be compared with the flint assemblage recovered from Hacklinge Holes, situated on a chalk ridge some 3km to the west. Here a rubbish pit containing pottery of c. 900-800 BC produced a very similar flint industry (Parfitt 1983), characterised by an indiscriminate use of flint pieces and a 'smash and grab' technique (Halliwell forthcoming). Such a late dating for the Dickson' s Corner material does not, however, allow for the presence of a classic Neolithic polished axe fragment; but this particular piece looks very much out of place with the rest of the material from the site and may not be related to the main industry. The possibility of the continued use of flint implements into the Romano- 140 A ROMAN OCCUPATION SITE AT DlCKSON'S CORNER, WORTH British period has often been discussed but conclusive evidence to demonstrate the point beyond doubt is very difficult to find (Young and Humphrey 1999). On the evidence of the later prehistoric material, a technologically very poor Romano-British industry may be anticipated and the bulk of the Dickson 's Corner material could fit very well into such a late group. From the available evidence, however, the date of the Dickson's Corner flint-work must remain uncertain. Future work in the Sandhills may yield clearer evidence for prehistoric occupation but at the moment it seems unwise to accept the present material as providing definite evidence for this. Animal Bone based on a report by Kevin Reilly A total of 189 animal and fish bone fragments were recovered and a detailed report (Reilly 1984) concerning these is held with the site archive. Just under half the bones are identifiable to species (Table 6). A few bones had been gnawed by dogs and some are burnt. The report concludes that:- It is likely that the animal husbandry practiced at this site was mixed and unintensive, especially regarding ox, sheep and goat, which almost certainly were killed and possibly bred at the settlement. Pig may have been brought to the site, while the evidence for horse is too slight to make any conclusions. A supplement to the diet was provided by red and roe deer and cod fish. The apparently large percentage of wild game may indicate 'an environment where agriculture and stock rearing was difficult' (Grant 1981, 207); however, this may also relate to availability. TABLE 6. SPECIES REPRESENTATION (total bone fragment count and minimum number of individuals - MNI) Species No. % MNI Ox, Bos taurus 34 38.2 3 Sheep, Ovis aries 3 3.4 I Goat, Capra hircus I I. I I Sheep/Goat. Ovis ariesl 27 30.3 3 Cavra hircus Pig, Sus scrofa 6 6.7 2 Horse, Equus sp. l 1.1 l Red deer, Cervus elaphus 4 4.5 l Roe deer, l 1.1 l Capreolus capreolus Cod, Gadus morhua 12 13.5 I Total identifiable 89 47. l - Unid. Mammal 89 - - Unid. Fish 11 - - Total unidentifiable 100 52.9 - GRAND TOTAL 189 - - 141 KEITH PARFTIT Marine Shell by Joan Briggs Only thirty-five marine shells were recovered from the site. Nearly half of these were unstratified, the rest being scattered in small numbers throughout the excavated deposits, with no particular concentrations. Oyster shells were the most common species represented, twenty-eight individual valves being found. Seven of these came from the domestic dump layer adjacent to Pebble Area E [DC 46]. Only four of the oyster shells are intact with their exterior markings visible; the remainder show varying degrees of wear. There are also five whelk shells (one, only a fragment) and three of cockle. The cockle shells are all in good condition. Considering the coastal location of the site it is surprising that so few marine shells were discovered. All the shells that were found are in a reasonable state and there is no reason to believe that many others have decayed without trace. DATING AND DISCUSSION The dating of the site at Dickson's Corner must be based on the pottery, with the 'Nauheim derivative' brooch (Fig. 10.4) providing an additional datable piece. The earliest deposits examined appear to be the carbon layer [DC 6], exposed in the I 978 Ditch (Assemblage I) and the primary hearth [DC 44] located in the SE Area Excavation (Assemblage 8). Pottery recovered from these deposits indicates that they date to around the middle of the first century AD, probably to just after the Roman invasion but conceivably a little before. The building represented by Pebble Area E was perhaps first erected in the late first century AD but certainly remained in use until the later second. From its position in the site sequence, the clay floored structure [DC 43] in the SE Area Excavation must have been in use during the late first century before being covered by a layer of wind-blown sand [DC 42]. The top of this sand had become mixed with domestic rubbish, including much pottery ranging in date from the late first to the second century (Assemblage 9), before Pebble Area C was laid across it, probably about AD 170. This pebbling, perhaps supporting a small building, probably continued in use until the earlier third century. No later Roman deposits or structures were located anywhere on the site. Overall, it would seem that the site was occupied from c. AD 50 until c. 225, or a little after. There is perhaps just sufficient evidence to indicate a pre-Conquest origin for the settlement; either way, the site was certainly occupied by soon after the Roman Conquest. Why it was abandoned during the early third century remains unclear • changes in local coastal topography is one obvious possibility (see below). A coin hoard of Tetricus, c. AD 271 (Roach Smith 1882; Dowker 1900, 111-2; Chapman 1921, 118), located in the dunes 142 A ROMAN OCCUPATION SITE AT DICKSON'S CORNER, WORTH opposite Blackhorse Wall, about 1.5km south of Dickson's Corner, however, suggests that the Sandhills were still occupied, or at least occasionally visited, in the later third century. The coastal location of the site inevitably raises questions concerning its relationship to the contemporary Roman shore-line but details of past sea-level changes are still poorly understood around much of the Kentish coast. The excavation of a Roman quayside at Dover, about 16km south of Dickson 's Corner, has provided some important information and suggests that high-tide in the Dour estuary during the late first-early second century 'cannot normally have been much higher than 4ft [ 1.22m] above OD ... ' (Rahtz 1958, 116). In today's harbour it can rise to 3.63m above OD. It has been calculated elsewhere that upon an open sea-shore a safety margin of about 1.80m would need to be allowed between the highest tides and lowest buildings of a site (the Minimum Occupation Level; Waddelove and Waddelove 1990, 254). Based on the evidence from Roman Dover, a Minimum Occupation Level of about 3m above OD may thus be suggested for the settlement at Dickson' s Corner. The majority of the Roman deposits and structures found on the site lie 2.30-3.S0m above OD which implies that the settlement stood in a region that was at threat from the sea, especially when there were storm surges. Another, unknown, factor, however, must have been the actual distance of the site from the shore-line (see Fig. 13). Although there was no clear archaeological or sedimentological evidence that the site had ever been inundated by the sea, in the recent past such events have occurred on a large scale in 1953 and again in 1978. In earlier centuries flooding may well have been on a more regular basis. Nevertheless, a light scatter of medieval pot-sherds recovered from the upper sands sealing the Roman deposits indicates subsequent activity in the area, during the eleventh - thirteenth centuries. Work near the Chequers Inn, about 1km south of Dickson's Corner, has indicated the presence of a settlement there during the medieval period (perhaps the documented Spruckelham or Langhauke; Tomaszewski 1978; 1979a; 1979b). Despite the fairly limited nature of the investigations, the overall character of the site at Dickson' s Corner seems reasonably clear. It appears to represent a native settlement occupied throughout the first and second centuries and into the third century AD. The earliest occupation may have occurred just before the Roman invasion of AD 43. At least one wooden building, probably a small dwelling, was located and further remains, including floors and hearths, may have formed parts of other structures. It seems most unlikely that any masonry buildings ever existed here. Pits and ditches, generally 143 KEITH PARFITT Fig. 13 Imaginative reconstruction of the Dick son's Corner site, c. AD I 00 (by Ben Stocker) common on Roman rural sites, appear to be largely absent and the settlement was probably unenclosed (Fig. 13 ). The isolated and exposed location of the site could suggest that occupation was seasonal rather than permanent. In several areas near-sterile sand layers [DC 4, 42, etc.), seemingly of wind-blown origin, were found separating Roman layers. Whether these deposits represent periods of complete abandonment or merely relate to localised shifts in the focus of the settlement, remains uncertain. Today, strong winds can cause the accumulation of up to 0.50m of sand across small areas in a day. Nevertheless, the pottery dating of the sequence of deposits recorded in the 1978 ditch and the SE Area Excavation, suggests that there were significant time-gaps between phases of occupation at these specific spots. The coastal location (Figs I, 2 and 13) suggests that the sea was a major element in the everyday lives of the inhabitants of the site and there is some limited evidence for sea-side industries. A small quantity of cod fish bones and a number of lead objects, probably weights from fishing nets, imply that fishing was undertaken, whilst the presence of marine shells suggests that shell-fish were also 144 A ROMAN OCCUPATION SITE AT DICKSON'S CORNER, WORTH collected and eaten, at least occasionally. Re-fired pottery recovered from the rubbish dump adjacent to Pebble Area E may be associated with small-scale salt production but if so, this could have only served the needs of the settlement itself. The general absence of ChaffTempered Ware, also perhaps connected with salt production, is possibly significant in this context. Quernstone fragments imply that grain was processed on the site as a regular domestic activity. It seems unlikely that cereals could have been grown locally in the dunes and presumably corn was brought to the site from fields on the higher land some distance away. Bones suggest that animal husbandry was practised. Ox, sheep, goats and pigs could have been bred here but may have been brought in. Red and roe deer, presumably hunted on the adjacent downlands, supplemented the diet, whilst fish may have been rather more commonly consumed on the site than the available bone evidence would suggest. Without doubt, the most significant point concerning the settlement at Dickson's Corner is its very location, for the site lies in an area long held by historians and archaeologists to have been under the sea in Roman times. From earlier finds, however, it is clear that this was not the only Roman settlement in the Sandhills north of Deal (Parfitt 1982). A pipe-trench cut through the region in 1992-4 failed to reveal anything of specific interest (Hearne, Perkins and Andrews I 995, 267-274) but other Roman occupation sites (and perhaps prehistoric ones, too) probably still await discovery here. The various Roman finds made in the Sandhills provide valuable dating evidence for the chronological development of the coastal sand spit. The most northerly site so far discovered, that at Dickson' s Corner, was occupied from around the mid-first century AD and from this it follows that the spit had moved at least this far north by the time of the Roman Conquest. If the evidence of the prehistoric flints recovered is taken at face value, however, it may be inferred that the spit had actually reached this point a thousand years or more before. The growth of the Deal sand spit out into the Wantsum was, then, well established by the beginning of the Roman period but the state of the present-day marshland (the Lydden Valley) further to the west, at this time, is less clear. Osborne White ( 1928) suggested that the silting of this part of the Wantsum was already well advanced by Roman times and Hawkes (1968) was in agreement with this. The discovery of Neolithic-Bronze Age flints sealed below the alluvium of the Lydden Valley indicates that substantial areas of land were habitable in prehistoric times (Halliwell, 1981; Halliwell and Parfitt 1985). Long (1992) has recently demonstrated the existence of a deeply buried prehistoric river channel running across this area. This 145 KEITH PARFITT was in-filled between the Mesolithic and early Bronze Age, when the progressively rising sea level was still significantly lower than today. Although the work at Dickson's Corner has provided some significant new information, there clearly remains much work to be done concerning the complex geological evolution of the Wantsum Channel and this part of the Kent coast before any detailed history can be set out. ACKOWLEDGEMENTS Thanks are due firstly to the respective land-owners, Wing Commander Nichollson and Mr Richard Daw, who readily allowed access to the site. Of the diggers who conducted the excavations, Mrs Joan Briggs, Mrs Shirley Creasey, Mrs Jill Bowers and Messrs Michael Halliwell, Geoff Halliwell, Edward James, Charles Burch, Howard Jones, Pat Mongovan and the late John Bray deserve a special mention for their hard work over many weekends. In addition, Charles Burch made available a heated greenhouse in which the finds could be processed. The site plans and sections have kindly been prepared for publication by Miss Jo Bacon and Barry Corke. Malcolm Lyne has drawn, and produced the report on, the coarse pottery, after initial work by Nigel Macpherson-Grant. Kevin Reilly has studied the animal and fish bone recovered. Ben Stocker has prepared an imaginative reconstruction drawing of the site (Fig. 13). BIBLIOGRAPHY Armour-Chelu, M. and Clutton-Brock, I 985, 'Notes on the evidence of cattle as draught animals at Etton', in F. Pryor, C. French and M. Taylor, 'An interim report on excavations at Etton, Maxey, Cambridgeshire 1982- 1984', Antiquaries Journal, 65, Part II, 297-302. Baker, J. and Broth well, D., I 980, Animal Diseases in Archaeology {London). Bell, M., 1977, Excavations at Bishopstone (Sussex Arch. Soc., Lewes). Blockley, K., Blockley, M., Blockley, P., Frere, S. S. and Stow, S., 1995, Excavations in the Marlowe Car Park and surrounding areas (The Archaeology of Canterbury Vol. 5). Bushe-Fox, J.P., 1926, First Report on the Excavation of the Roman Fort at Richborough, Kent (Rep. Res. Comm. Soc. Antiq. London 6), London. Bushe-Fox, J. P., 1949 Fourth Report on the Excavations of the Roman Fort at Richborough, Kent (Rep. Res. Comm. Soc. Antiq. London 16), Oxford. Chapman, H. S., 1921, The Story of Dola. Julius Caesar's Landing Place (London). 146 A ROMAN OCCUPATION SITE AT DICKSON'S CORNER, WORTH Clarke, A., 1964, O.S. Record Card, TR35 NE I. Davies, B., Richardson, B. and Tomber, R., 1994, The archaeology of Roman London Volume 5. A dated corpus of early Roman pottery from the City of London (C.B.A. Res. Rep. 98). Dowker, G., 1900, 'Deal and its Environs'. Archaeologia Cantiana, 24, I 08- 121. Fordham, S. J. and Green, R. D., 1973, Soils in Kent II. Sheet TR35 (Deal) (Soil Survey record No. 15). Geoffroy, J. F.,Calonne, E. and Thoquenne, V ., 1998, 'Etude techno-typologique de la ceramique decouverte sur le site de la ferme indigene d'HeninBeaumont (Pas-de-Calais), campagne 1994', in M. Tuffreau-Libre and A. Jacques (eds.), La Ceramique Precoce en Gaule Belgique et dans les Regions Voisines: de la poterie Gauloise a la ceramique Gallo-Romaine (Nord-Ouest Archaeologie No. 9), 77-100. Grant, A., 1981, 'The significance of deer remains at occupation sites of the Iron Age to the Anglo-Saxon period', in M. K. Jones and G. Dimbleby (eds), The Environment of Man: the Iron Age to the Anglo-Saxon period, 205 - 213 (BAR Brit. Ser. 87). Halliwell, G., I 98 I, 'Flint Artifacts from the Sandwich-Deal Marshes', Kent Arch. Rev., 65, 113-116. Halliwell, G. and Parfitt, K., 1985, 'The Prehistoric Land Surface in the Lydden Valley', Kent Arch. Rev., 82, 39-43. Halliwell, G., forthcoming, 'The Flint-work·, in Parfitt, K .. 'Two Pits at Hacklinge Holes, Worth'. Hawkes, S., 1968, 'Richborough - the Physical Geography', in B. W. Cunliffe (ed.) Fifth Report on the Excavations of the Roman Fort at Richborough, Kent. (Rep. Res. Comm. Soc. Antiq. London 23), London. Hearne, C., Perkins, D. R. J. and Andrews P., 1995, 'The Sandwich Bay Wastewater Treatment Scheme Archaeological Project, 1992-1994', Archaeologia Cantiana, 115, 239-354. Long, A. J ., 1992, 'Coastal responses to changes in sea-level in the East Ken I Fens and southeast England, UK, over the last 7500 years', Proc. Geol. Assoc., 103, 187-199. Keller, P. T., 1989, 'Quern Production at Folkestone, South-East Kent: An Interim Note', Britannia, 20, 193-200. Macpherson-Grant, N., I 980, 'Chaff tempered Ware', Kent Arch. Rev., 61. 2-4. Marsh, G., 1978, 'Early Second Century Fine Wares in the London Area'. in P. Arthur and G. Marsh (eds.), Early Fine Wares in Roman Britain, 119-223 (BAR Brit. Ser. 57). Monaghan, J., 1987, Upchurch and Thameside Roman Pottery. A ceramic typology for northem Kent.first to third centuries A.D. (BAR Brit. Ser. 173). Orton, C., 1975, 'Quantitative pottery studies: some progress, problems and prospects', Science and Archaeology, I 6, 30-5. Osborne White, H., I 928, The Geology of the country, near Ramsgate and Dover (Mem. Geol. Survey). Parfitt, K., 1980, 'A Roman Occupation Site at Worth'. Kent Arch. Rev., 62, 46-47. 147 KEITH PARFITT Parfitt, K., 1982, 'Roman finds from the Sandhills, North of Deal', Kent Arch. Rev., 70, 225-227. Parfitt, K., I 983, 'Reports from Local Secretaries and Groups: Hacklinge Holes', Archaeologia Cantiana, 99, 290. Peacock, D. P. S., 1977, 'Pompeian Red ware', in D. P. S. Peacock (ed.), Pottery and early commerce. Characterisation and trade in Roman and later ceramics. Pollard, R. J., 1988, The Roman Pottery of Kent (K.A.S. Mono. Series 5, Maidstone). Pollard, R. J., 1995, 'The Pottery: Mid to Late Roman Periods', in Blockley, K. el al., 690-736. Rahtz, P. A., 1958, 'Dover: Stembrook and St Martin-le-Grand, 1956', Archaeologia Cantiana, 72, 111-137. Reilly, K., 1984, 'Dickson's Corner: Animal bone report from a RomanoBritish Farmstead' (archive report prepared for Dover Arch. Group). Roach Smith, C., I 882, 'Retrospective Observations respecting a Hoard of Roman coins found in the Sandhills. near Deal', Archaeologia Cantiana, 64, 69-81. Robinson, A. H. W. and Cloet, R. L., 1953, 'Coastal Evolution in Sandwich Bay', Proc. Geol. Assoc., 64 (Part 2), 69-81. Shephard-Thorn, E. R., 1988, Geology of the Country around Ramsgate and Dover (Mem. Geol. Survey). Tomaszewski, N., 1978, 'Medieval Pottery from Deal', Kent Arch. Rev., 54, 93-94. Tomaszewski, N., 1979a, 'Spruckelham: near Deal. A lost ham?', Kem Arch. Rev., 56, 142-144. Tomaszewski, N., 1979b, 'Investigations and Excavations during the Year: Deal', Archaeo/ogia Cantiana, 95, 307. Waddelove, A. C. and Waddelove, E., 1990, 'Archaeology and research into sea-level during the Roman era: Towards a methodology based on Highest Astronomical Tide', Britannia, 21, 253-266. Webster, J., 1975, 'Objects of Lead', in B. W. Cunliffe, Excavations at Por1chester Castle Vol. I, 232-3. (Rep. Res. Comm. Soc. Antiq. London 32) London. Wilson, M., 1983, 'The Pottery', in S. S. Frere and S. Stow, Excavations in the St George's Street and Burgate Street areas (The Archaeology of Canterbury Vol. 7), 192-300. Wilson, M., 199S, 'Pottery from C.E.C. sites', in K. Blockley et al. 199S, 682-89. Young, R. and Humphrey, J ., 1999, 'Flint use in England after the Bronze Age. Time for a re-evaluation', Proc. Prehist. Soc., 6S, 231-242. 148