An Early Romano-British Salt-Working Site at Scotney Court

AN EARLY ROMANO-BRITISH SALT-WORKING SITE AT SCOTNEY COURT L. BARBER INTRODUCTION The site at Brett's Scotney Court gravel pit was located some 4 km to the south-west of Lydd at N.G.R. TR 014194 at an average elevation of 1 to 2.5 m O.D. (Fig. 1). The extraction has now ceased and the location of the excavation is marked by a water-filled gravel pit. The main flint gravel deposits at Scotney Court for the most part lay below the fine-grained sediments and below the water-level in the worked pits. The gravel rises towards the south-east and is exposed on the surface at The Forelands, which lies within the Ministry of Defence ranges. Finds of Roman pottery and briquetage were initially made at Scotney Court during 1980 by Mr J Moody whilst mechanically stripping the topsoil prior to gravel extraction.1 The find-spot, which was not precisely recorded, was not examined by archaeologists, but it was suggested that the remains lay in a ditch. The pottery was all of Claudio-Neronian date which correlates well with that found in the more recent excavations. During May 1989, Mark Gardiner, then of the Field Archaeology Unit, University College London, visited the site with Jill Eddison of the Romney Marsh Research Trust and noted evidence of an extensive archaeological deposit in an exposure of the gravel pit. A narrow causeway between two water-filled pits had been created by the extracting of gravel either side (Fig. 1). On the south-eastern face of the causeway, briquetage, pottery and charcoal were traced for a distance of about 100 m. Further gravel extraction which was to 1 B. Philp and J. Willson. 'Roman site at Scotney Court, Lydd', Kent Archaeol. Rev., 68, 1984, 156-61. 327 L. BARBER KENT Scotney Court --'-T SUSSEX Scotney Court Kent ^ ^ \ S c o t n e y Court Works Forelands Quarry Fig. I. Site location 328 AN EARLY ROMANO-BRITISH SALT-WORKING SITE AT SCOTNEY COURT remove the causeway threatened this exposure and during Easter 1990 Mark Gardiner recorded the deposits in section (Fig. 2, points I to III and V) and undertook a small excavation (Fig. 2, Trench IV) in order to obtain material for dating purposes. Following the Easter work it was decided to undertake more extensive investigation of the site. Brett Gravel kindly provided a 360° tracked excavator to strip the topsoil off parts of the remaining deposits which enabled a small team from the Field Archaeology Unit, University College London, assisted by a handful of local volunteers, to undertake a three-week long excavation in December 1990. Only four trenches were opened, A-D (Fig. 2). All were of limited size as the causeway had to be kept clear for the movement of heavy plant which supplied the drag-line on the end of the causeway with fuel. All stages of work were funded by the Romney Marsh Research Trust. A full account of the results of the excavations and finds analysis is given in the Archive report (Barber 1996). The current paper represents a concise account of the work. THE EXCAVATED FEATURES Trench A Trench A consisted of an irregular area measuring in excess of 10 x 20 m. (Fig. 2). The trench was located in order to investigate a possible salt evaporation pan (Context 73) and an apparently domestic pit (Context 49) which had been observed in the section of the quarry. The salt pan did not extend into the area of Trench A. It therefore seems probable that the surviving portions of it visible in the quarry section were the last remains of the northern edge and that most had already been removed by quarrying to the south. The section of the quarry was cleaned and recorded as a result. This showed the feature to be large but shallow, with a maximum depth of 500 mm. Although the eastern edge was located the general indistinct edges made it impossible to trace its full extent to the west with certainty. Three fills were apparent in the section (74, 58 and 81) but none contained any datable material. The pit which was noted to the east was found to extend into the area of Trench A (Context 49, Fig. 2). Although the north-eastern side of the feature appeared to be truncated by a later drainage ditch (not illustrated) the overall plan was kidney-shaped with steepsloping sides and a flat base. A number of fills were apparent (57, 46, 42 and 39) within the pit. Many of these produced evidence of domestic activity in the form of first-century pottery and fish bone. 329 ouV,S4° 0150/tBSt Quarry Quarry Edge Shingle exposed \ in Section TRENCH D lc-122 11&BPC. —Jo. I0U5/1937 Quarry TRENCH C Quarry 3i>-^—4^C"- Easter Trench **-—^ IV ^ TRENCH B West extent \ ot73 > TRENCH A 1 — k DTestPiT- • - j \ \ East extent ot 7 3 - \ \. 1 _ . ^ . _ 4 ! * i Quarry 20m _j i I 0154/iW Fig. 2. Location of Trenches AN EARLY ROMANO-BRITISH SALT-WORKING SITE AT SCOTNEY COURT The latest fill of Pit 49 (39) lay immediately below the 400-500 mm. thick topsoil layer (38). With the exception of a single first-century post-hole or small pit (Context 43) no other features of archaeological interest were noted cutting the natural clay within the trench. Trench B Of all the trenches Trench B exhibited the most difficult stratigraphy to interpret (Fig. 3, Section F). The lowest exposed layer within the trench was a yellow brown silt clay containing no archaeological material (75). Above this was a deposit of dark brown silt clay with very occasional briquetage flecks (76). The briquetage content gradually increased toward the west where 76 darkened and merged into a dark grey silt clay layer containing abundant briquetage and charcoal (77). Context 77 can be seen as the eastern edge of the extensive briquetage layer seen further west (Context 83 in Trench C, see below). The way in which this layer gradually merged into 76 with a lightening in colour and lessening of briquetage content, along with the occurrence of dark feature-like pockets below the layer suggest it had been partially remodelled by water action. Below 77 at the western end of the trench were four further layers (52, 53, 54 and 55). Stratigraphically 52, 54 and 55 (not illustrated) are similar to 75; however, they were different in character probably as a result of staining/mixing with 77 which lay immediately above. Cutting 54 and 55 were a number of distinct but often irregular features, some of which may well have been the result of water scouring and subsequent deposition. Only two appear in the illustrated section (64 and 69). Sealing the briquetage layer was a thick disturbed overburden (78-80). To the south of the trench, in the quarry face, a circular shallow cut (Context 71, Fig. 2 only) was noted. Its fill (72), contained flint shingle, briquetage and charcoal fragments along with a little firstcentury pottery. Trench C The 'natural' clay (84) was cut by a number of features (Figs. 2 and 3, Section H). The most easterly appeared to be a north-south ditch (86) which was visible in the quarry-face to the south but did not appear in the quarry face to the north. The ditch was fairly shallow but had an irregular profile. Its fills (93, 91, 88 and 87) all contained briquetage pieces. A gully was present cutting 87 and 88, and it is possible this had resulted from water draining along the ditch. The nature of the gully's fill (85), suggested it to have derived from the briquetage layer above (83). 331 Trench B Section F p j w r r T ^ V T W F r n; tJH^ililL' |T''^^T1T7ri3WTTn>rTTTrr^rTTTTrn^T^^ l ' l ' I T I ' l ' | i | i | i | ' i nrrum^r^n^Trr^TVVr'r'''' re. _Li* J JjJg, -k jtji&.u./ 52 70 Trench C Section H TTnTrrjrpjrrrprrr^^ •^tf, Trench D Section J 111 113 '—' lit siiroi^^ MTIWITHITITHTIT^ <123 100 Fig. 3. Sections: Trenches B, C and D 332 • AN EARLY ROMANO-BRITISH SALT-WORKING SITE AT SCOTNEY COURT To the west was a shallow irregular cut (89); two small oval pits or post-holes (95 and 97) and a smaller post-hole (104). The upper fills of all these features were similar to the briquetage layer above (83). A narrow, roughly north-south gully with steep sides and rounded base (102) lay further west. The gully fill (103) was similar to those of the other features. At the western end of the trench was a large but shallow cut (99). Above this was a layer with small quantities of briquetage pieces (100). This layer not only filled the depression caused by 99 but continued either side and the exact nature of this layer/fill remains unascertained. To the south of Trench C was a small pit/post-hole in the quarry edge (142). This was in a similar stratigraphic position to the features within the trench and consisted of a steep-sided circular cut measuring 550 mm. in diameter by 500 mm. deep. All features within Trench C appear to date to the first century and were sealed by a thick layer of compact dark brown silt clay containing flint pebbles to 50 mm. and 10 per cent briquetage and charcoal flecks (83). This layer of salt-working debris was extensive, appearing in both Trenches B and D as well as the whole length of the southern quarry face for the same distance. However, it was not noted on the northern side of the peninsular. This briquetage layer was sealed by a thick layer of topsoil/subsoil (82). Trench D This trench proved to be the most significant as it contained a complex sequence of both natural and archaeological deposits (Figs. 2 and 3, Section J). The natural clay (108) was cut by a number of features. The earliest was a steep-sided ditch (118). Its fill (119) consisted of a homogenous light grey clay with 90 per cent flint shingle averaging 10 mm. in size. No initial silting was encountered in the ditch suggesting it had been filled with shingle, derived from Context 111, soon after being dug. The shingle layer (Context 111), was part of a tongue of shingle which had apparently been shot landward, from the south-west, by a storm event. It ran across the trench at an angle and tapered down toward the north. The shingle consisted of 90 per cent ill-sorted flint pebbles, ranging in size from 10 to 60 mm. (though most were between 20 to 40 mm.), set in a light grey clay matrix. Context 122 was a shallow ditch at the western end of the trench. The lower fill (141) contained some briquetage and charcoal flecks and very sparse flint pebbles. The upper fill (123) contained far more flint pebbles suggesting this fill may have been deposited after the shingle tongue. Although no pottery was recovered from the lower 333 L. BARBER fill, Context 123 produced 12 first-century sherds. A similar date and stratigraphic position is assumed for the isolated post-hole (120), located toward the eastern end of the trench. Cutting 111 in section was a shallow scoop (115) and a small pit (113). Neither feature contained any dating evidence. Above 113, and overlaying the shingle 111, was a dark grey brown silt clay layer (112). This layer is likely to be of the same origin as 107 to the east, which was similar, but contained fewer pebbles and briquetage, and 137 to the west. Layer 137 was originally thought to be a fill in a large shallow cut (136); however, it now seems likely that this cut is in fact the natural western edge of the shingle tongue. Whatever the case, it is clear that salt-making was continuing, with the waste material being dumped on top of the natural clay and shingle. Two parallel ditches were subsequently cut across the area (117 and 110). Ditch 117 cut through the briquetage and shingle layers (112 and 111) and consisted of a steep-sided flat-bottomed cut containing a number of fills (Contexts 134, 132, 133, 131, 130, 129 and 128). The lack of flint pebbles in the primary fill (134) was surprising considering the ditch cut 111 and it is possible 134 was a deliberate infill or sudden slump of natural clay (108). The majority of the remaining fills contained much higher percentages of shingle along with briquetage fragments. Unfortunately no datable material was located in any of the fills of ditch 117; however, it is in a similar stratigraphic position to Ditch 110 (see below) and may be of similar date. Ditch 110 had a similar profile to that of 117 but contained a less complex sequence of fills (140, 138, 139 and 109). The earliest fill (140), a dark brown organic silt, contained very occasional flint pebbles and fragments of decayed reeds. This merged with 138 which had been heavily derived from the shingle to the west. Most of the pottery in 138 is of abraded first-century types; however, four sherds of unabraded beaker were also located which date to the second or early third centuries. Both ditches 110 and 117 were sealed below the topsoil/subsoil (105). THE EXCAVATED FINDS The Pottery (incorporating comments by Sue Hamilton, Nigel MacPherson-Grant and Valery Rigby). Introduction The excavations produced a relatively small quantity of pottery (288 334 AN EARLY ROMANO-BRITISH SALT-WORKING SITE AT SCOTNEY COURT sherds weighing 3,500 grams from 25 contexts). Although no large groups are present, the pottery is an important assemblage as it is the first of the period from the Marsh and for this reason is treated in some detail in the archive report. Whether this is a typical assemblage or one related to the salt-making industry cannot be ascertained until further sites of the period have been studied on the Marsh. The pottery from Scotney Court is virtually all of first-century A.D. date and probably spans the period A.D. 25 to 100. This closely correlates with the pottery found at Scotney Court in I980.2 The pottery from the adjacent site at Lydd Quarry appears to be of a slightly later date (later first to second century) and as such provides a useful continuation of the ceramic sequence.3 All pottery was divided into fabric groups and quantified by context. This information is recorded on pottery summary sheets which form part of the archive. All diagnostic sherds have been illustrated for this paper (Figs. 4 and 5). The Fabric Groups Two main fabric groupings are present in the assemblage: grogtempered and sand-tempered wares. The fabrics from the 1990 excavations broadly agree with those of the 1980 finds. Grog-Tempered Wares The hand-made grog-tempered fabrics (Fabrics la to lg) are by far the largest group representing 85.3 per cent of the overall assemblage by weight. All contain grog inclusions of varying sizes and quantities but some have other inclusions such as flint. Unfortunately, not enough material is present to be sure if there is any chronological reason for the various sub-groups. Where rims are present (Figs. 4 and 5) they appear to be of Late Iron Age to early Romano-British types. However, these predominantly represent the indigenous Iron Age traditions with relatively minor influence from post-Conquest ceramics. How long these forms lasted in such a remote area is difficult to ascertain without the presence of more diagnostic 'imported' material. All percentages are for weight and are based on the whole assemblage. Full fabric descriptions are given in the archive report. 2 Philp and Willson 1984. (ibid.) 3 L. Barber, forthcoming, The Roman Pottery from excavations at Lydd Quarry. 335 L. BARBER ; .".• .^ t 7 \ ( : Black PHcn/Paint Fig. 4. Pottery nos. 1-15. Scale 1:4 Trench IV, Layer 22, 1-5: Trench A, Pit 49, 6-15 336 AN EARLY ROMANO-BRITISH SALT-WORKING SITE AT SCOTNEY COURT ? - • •) ? A / / > \ ••.'•'•M\i.':\% mm, f A Black Pilch S*« Fig. 5. Pottery nos. 16-32. Scale 1:4 16-19: Pit 49. 20: Post-hole 43. 21: Layer 52. 22-23: Layer 83. 24-26: Ditch 86. 27: Cut 99. 28-29: Ditch 122. 30-31: Ditch 110. 32-33: Pit 142. 337 L. BARBER la Oxidised sparse fine to medium grog-tempered (Al per cent). Figs. 4 and 5. Nos. 7-16, 21, 23, 32 and 33. 1 b Unoxidised sparse to moderate fine light grey grog-tempered (7.4 per cent). Nos. 6, 22 and 28. 1 c Unoxidised abundant medium light grey grog-tempered (19.6 per cent). This fabric is a coarser variant of lb. Nos. 24, 26, 27 and 29. Id Unoxidised light grog and iron-ore tempered (1.3 per cent). None illustrated, le Oxidised Sparse Grog and very occasional calcined flint tempered (2.7 per cent). No. 25 1 f Unoxidised sparse to moderate light grey grog with moderate black and white flint grits (8.1 per cent) Nos. 2, 3 and 4. 1 g Unoxidised grog and sparse iron-ore tempered (5.2 per cent). variant of fabric Id but with slightly more abundant grog and iron-ore inclusions. None illustrated. Sand-Tempered Wares The two related sub-groups of sandy wares make up 7.7 per cent of the overall assemblage by weight. These wares were recognised in the pottery found in 1980.4 Most appear to be typical Kentish fine sandy wares of Late Belgic to early Romano-British origin. 2a Oxidised sandy ware (A.A per cent) Nos. 17 and 18. 2b Unoxidised sandy ware (3.3 per cent) This is a variable group and includes a sherd of BB2 from Ditch 110 (Fig. 5, No. 31) Nos. 20 and 31. Other Fabrics 3 Flint tempered (4.9 per cent) The only form discernible is the bucket-like urn of Mid Iron Age-type (as Group lc). It is likely, however, this form may have continued in use into the first century A.D. at Scotney Court. Nos. 1 and 5. 4 Sand and shell tempered (0.7 per cent). None illustrated. 5 Abundant shell tempered (0.2 per cent). None illustrated. 6 Shell and flint tempered (0.2 per cent). None Illustrated. 4 Philp and Willson, 1984. (ibid.) 338 AN EARLY ROMANO-BRITISH SALT-WORKING SITE AT SCOTNEY COURT 7 Fine greyware with grog and some flint (0.1 per cent). None Illustrated. 8 Moderate sandy fine ware (0.2 per cent). None Illustrated. 9 Fine buff powdery ware with sparse red grog-temper (0.2 per cent). None Illustrated. 10 Colchester Colour-coat ware (0.5 per cent). No. 30. The Briquetage Introduction The excavations produced 686 pieces of briquetage, weighing 7,602 grams, from 35 contexts. This total represents a small but representative proportion of the material that was present on site as a rigorous sampling strategy had to be undertaken during the excavations. As is common with low-fired briquetage from other sites, most of the material consists of small fragments, the majority of which are badly abraded.5 Much of the material is, therefore, undiagnostic of form. Some pieces are harder fired, usually to a dull grey or purple, although this is likely to be the result of heating during salt production than representing initial firing. Six 'fabric types' were identified and are fully described and quantified in the archive report. All fabric types appear in contexts associated with first-century pottery and often appear together in individual contexts. It is, therefore, fairly certain that all the briquetage from the excavations is of first-century date and used contemporaneously. The precise uses and complete form of many of the briquetage vessels are still uncertain, although there does appear to be some fabric types used predominantly for certain types of briquetage. The Fabric Types Bl Abundant sub-rounded multicoloured flint grits. (Ill pieces weighing 983 g.) All sherds are from thin walled (5-14 mm.) vessels with simple rims. (Fig. 6, nos. 1, 2, 4 and 5) B2 Abundant sub-rounded to angular calcined flint grits. (Al pieces weighing 261 g.) Most intensive study appears to have taken place in Essex. Colchester Archaeological Group. 1990. The Red Hills of Essex: Salt-making in Antiquity. Annual Bulletin No.30. 339 L. BARBER \ \ it 1 \i___l__) I %."j sf, bS_^ in Ge 8 9 K t 1 * < . _ - ^ / h ,«if.eTficrtfif l 0 12 13 • X 11 to M i - IS & 5cms Fig. 6. The Briquetage. Scale 1:4 1-5: Trench IV, Layer 22. 6: Layer 53. 7, 9, 11, 13 and 14: Ditch 86. 8: Ditch 102. 10 and 12: Unstrat. 15-17: Layer 53. 18: Layer 100. 340 AN EARLY ROMANO-BRITISH SALT-WORKING SITE AT SCOTNEY COURT Again, this fabric is exclusively used for thin-walled (6-14 mm.) vessels. No.3. B3 Moderate sub-rounded flint gritted. (39 pieces weighing 228 g.) As Fabric Bl but with less tempering. This fabric is only used for thin-walled vessels. None illustrated. B4 Moderate grog temper with occasional organic inclusions. (63 pieces weighing 619 g.) Some of the fragments are from thin-walled (6-11 mm.) vessels of square or rectangular form with plain right-angled bases. No. 7. B5 Moderate to abundant grog tempered. (416 pieces weighing 5336 g.) The fabric is rarely used for vessels; however, most of the pedestals, short rods and pinch-props are made in his fabric. Some fragments with rod impressions, possibly from an oven structure, are also present. Nos. 6 and 8-18. B6 Abundant organic tempered. (16 pieces weighing 175 g.) No forms are discernible; all the pieces consist of irregular lumps. None illustrated. The Forms As stated above forms are often difficult to ascertain. This is due to the small size of the fragments but is also a result of the unusual and irregular nature of the complete briquetage forms. Pedestals are represented by a number of diagnostic pieces (Nos. 10-11). These are of a standard type and are now well known from other sites.6 Similar pedestals have already been found at Scotney Court.7 All the present examples are in fabric B5 and are of rough cylindrical form. Although all sherds are too small to reconstruct any full profile, vessels appear to be of two main forms. Fabric Bl to B3 are exclusively used for crude splayed-sided cylindrical vessels with simple rims (Nos. 1-5). It is likely these were either used for evaporation or the storage of salt. Similar vessels have been found in Iron Age contexts in Essex.8 All vessels are low-fired with thin walls. One base sherd to a circular vessel in Fabric B5 was also located (No. 9). This is the only example of a circular vessel in this fabric, the 6 Colchester Archaeological Group 1990. (ibid.) 7 Philp and Willson, 1984. (ibid.) * Colchester Archaeological Group 1990. (ibid.) Fig.34 341 L. BARBER remaining six base fragments being from square or rectangular vessels (Nos. 6-8). These consist of thin-walled slab-built vessels with simple flat bases. Several body sherds of these vessels were also found but no large pieces are present making it impossible to judge their size. All sherds are in either fabrics B4 or B5. An evaporation or salt storage tray seems the most likely function of these vessels. Short props, consisting of crude and irregular clay 'sausages' are quite numerous in the assemblage (Nos. 12-15).9 Their form is irregular and precise function unknown. Most are in fabric B5. Pinch props, consisting of extremely irregular hand kneaded wedges of clay are another common form of briquetage, again, predominantly in fabric B5 (Nos. 16-18). It has been suggested these were used to wedge between the evaporating vessels, only becoming fired during the evaporation process.10 The Mammalian Assemblage Wendy Wood A total of 177 fragments of mammalian bone were recovered from the site. Full details of all the bone is recorded in the archive. The bone assemblage is a small one, and thus it is difficult to draw any accurate conclusions. The assemblage may indicate a trend towards preferred exploitation of Bos., supplemented by sheep/goat and pig, respectively. These species would be exploited as food resources, although the lack of immature bones may suggest that wool and milk also figured heavily in the economy. However, the interpretation of the site as an area of Romano-British salt production suggests that the majority of this assemblage became incorporated into deposits as food refuse. Some butchery was evident in the sample. Twelve specimens exhibited knife or chop marks to the shaft. These were confined to the mature bones of Bos. and Ovis. Fish Remains Brian Irving A soil sample from Pit 49 (Fill 46, Trench A) contained a total of 158 9 Colchester Archaeological Group 1990. (ibid.), 14, 10 Colchester Archaeological Group 1990. (ibid.), 13. 342 AN EARLY ROMANO-BRITISH SALT-WORKING SITE AT SCOTNEY COURT fish bone fragments of which 30 could be assigned to species level, the remaining material consisted of ribs and small cranial fragments which were very fragmentary. Of the 30 identifiable bones, 25 were identified as halibut (Hippoglossus htppoglossus), and are from a single individual which weighed approximately 2 kg. This estimate is based on comparison with modern reference material of known weight and length. The halibut material consisted of both vertebrae and cranial elements. Five bones of haddock consisted of three caudal vertebrae, a penultimate vertebrae and a post temporal fragment. Only the haddock material showed any evidence of butchery. One caudal vertebrae and a single penultimate vertebrae clearly showed that a filleting knife had been used to cut them laterally in a head to tail orientation. This type of butchery evidence suggests that the haddock was filleted whereby the knife cut which removes the flank, cuts or splits the tail vertebrae for drying and/or salting. Each flank is usually joined at the tail and is slung over a string for drying. Much of the salt produced from intertidal marshes during the Romano-British period was traded inland. The salt would mainly be used for food preservation so it would be reasonable to suggest that locally procured fish and salt were combined at a single site for the trade of stock fish both inland and abroad. Only two species have been identified from the assemblage, but both are important food fish which are still preserved today by salting. However, the butchery evidence, albeit from only two bones, points toward the site or its immediate environs being used as a fish-salting station. The Charcoal David Goode Introduction There were a total of six samples from the site. Of these, four were samples of charcoal while two contained wet/waterlogged wood. The range of species identified in the charcoal samples is summarised in Table 1. The identified species of waterlogged/wet wood is summarised in Table 2. A full report forms part of the archive. The pattern for the site indicates the prevalence of Quercus and Pinus. The large quantities of these two species do not allow for interpretation of analogous environments. However, as a whole the sample is dominated by the presence of dry soil/upland tree species. The location of the site - low lying land on a brackish marsh - would suggest that the dry-land species (Pinus, Betula, Carpinus, Corylus, Quercus type and Fagus) were deliberately brought to the site. The 343 L. BARBER TABLE 1: THE RANGE, MASS AND MASS PERCENTAGE OF IDENTIFIED WOOD CHARCOAL Species Pinus sylvestris Alnus glutinosa Betula type Carpnus betula Corylus avellana Populus/salix type Quercus type Total: 7 Species Mass 0.90 0.31 0.10 0.06 0.10 0.06 2.29 3.82 % of Total mass 23.5 8.1 2.6 1.6 2.6 1.6 60.0 100% TABLE 2: WOOD SPECIES IDENTIFIED IN WATERLOGGED/WET SAMPLES Species Alnus glutinosa Betula type Pinus sylvestris Fagus sylvatica Unidentifiable Total: 4 Species Total # 5 2 1 1 6 15 pieces remaining two species (Alnus, and Populus/salix type) are tolerant of wet soils and might easily be found on the periphery of a marsh. Results The range of species found in each of the three fills of Pit 49 (Trench A, Fills 42, 46 and 57) remains consistent throughout the pit. The basal fill of Pit 49 contained only fragments of Quercus type charcoal. The total mass of charcoal recovered from Pit 49 is 1.80 grams. The breakdown of mass percent for the entire pit reflects the trend in Context 42: Pinus sylvestris (50 per cent), Alnus glutinosa (17.2 per cent), Quercus type (12.5 per cent), Betula type and 344 AN EARLY ROMANO-BRITISH SALT-WORKING SITE AT SCOTNEY COURT Corylus avellana (5.2 per cent each), Carpinus betula and Populus/ salix type (3.4 per cent each), and unknown (2.8 per cent). Tree-ring analysis of an intact section of Pinus sylvestris branch shows that the branch represented approximately six years of growth, and that it was cut or broken from the tree trunk before the end of the summer growth season, i.e. before the manufacture of a new 'early wood'. Contexts 138 and 140 are both fills from Ditch 110 (Trench D). The range of species recovered from the two samples is completely dissimilar. This difference indicates that the ditch was receiving carbonised wood (large branches/stem wood) and fresh wood (twigs) during different phases. There was no evidence of cut-marks or working on any of the waterlogged specimens. Though few of the tree species indicated are tolerant of wet soil, the Populus/salix type and Alnus glutinosa can thrive in these conditions. Soil which could support the remainder of the species indicated is located within a few kilometres at the edge of the marsh. Plant Remains Pat Hinton Introduction Samples were submitted from the following contexts: Trench A - Pit 49, fills 42 and 57; Trench C, Layer 100 and Trench D - Ditch 110, fill 140. The results for charred and waterlogged items are tabled separately as they form two different categories. The charred cereals and characteristic arable weed seeds listed in Table 3 presumably became incorporated in the deposits by chance, but those preserved by waterlogging listed in Table 4 are predominantly plants which were likely to have been growing nearby. The numbers of some of the seeds have been estimated but the rush seeds in Context 140 are very numerous and probably present in thousands. All plants are represented by seeds (which word is used to include fruits, etc.) unless otherwise stated. Charred Seeds The charred cereals are all poorly preserved but sufficient remains to identify the barley grains as hulled but not to distinguish whether of 6 or 2-row form. Three grains in Context 42 are probably wheat, but are far too damaged to suggest species. Wild radish (Raphanus raphanistrum), often found as separated one-seeded segments of the siliqua (pod), is a weed usually of non-calcareous soils. Cleavers 345 L. BARBER (Galium aparine) and rye brome (Bromus secalinus) are typically weeds of autumn-sown crops. These charred remains probably were derived from crop-processing procedures. Waterlogged Seeds The much larger range of species in the waterlogged samples consists mainly of plants which grow in or beside water, in muddy places and damp grassland. Duckweeds (Lemna spp.) are floating plants which live on the water surface, while water crowfoot (Ranunculus Subg. Batrachium), fool's water cress (Apium nodiflorum), water plantain (Alisma plantago-aquatica), pondweed (Potamageton spp.) and club rushes (Schoenoplectus lacustris) are rooted in soil but grow through still or flowing open water of varying depths and rates of flow. Plants of pond and ditch margins and similar places are celeryleaved buttercup (Ranunculus sceleratus), water pepper (Polygonum hydropiper), marsh pennywort (Hydrocotyle vulgaris), the water drop- worts (Oenanthe spp.) water mint (Mentha aquatica), rushes (Juncus spp.), spike rushes (Eleocharis spp.) and the sedges (Carex sppl.). Buttercups, particularly creeping buttercup (Ranunculus repens), ragged robin (Lychnis flos-cuculi) and the thistles (Cirsium spp.) would have grown in damp grassland and many of the ditch and pond side plants might also have grown in such conditions. The weed and ruderal seeds such as fat hen (Chenopodium sp.) red dead-nettle (Lamium purpureum), field penny cress (Thlaspi arvense) seem less appropriate to the damp surroundings, as is the poppy (Papaver somniferum). The opium poppy has been cultivated for its oil-bearing seeds, but it becomes naturalised readily in lighter soils and so in this context may perhaps be regarded as a weed and, with the others, a chance intrusion. Henbane (Hyoscyamus niger), another plant of disturbed ground, is often found near the sea. Blackberry (Rubus Fruticosus) and elder (Sambucus nigra) could well have been gathered in nearby scrub or woodland but seeds of these fruits are also carried by birds. Fills 42 and 57 in Pit 49 differ in that the higher, 42, includes large proportions of water crowfoot, water plantain, duckweed and rush species, all plants of shallow, often gently flowing water. Only two seeds of water crowfoot were found in the lower fill, 57, where the dominant plant is common club-rush, which normally grows in somewhat deeper and usually still water. This plant is represented by only one seed in the upper fill. Water-dropwort seeds, only present in the Context 57 sample, form a significant proportion of the seeds in 346 AN EARLY ROMANO-BRITISH SALT-WORKING SITE AT SCOTNEY COURT TABLE 3: CHARRED PLANT REMAINS Context Sample vol. (litres) Hordeum vulgare L. Hulled barley cf. Triticum sp. Wheat Cerealia indet. Undetermined cereals Raphanus raphanistum L. Wild radish Galium aparine L. Cleavers Bromus cfsecalinus L. Rye brome POACEAE indet. grains rachis fragment grains fragments siliqua segments Pit 49 42 4.5 8 5 4 6 2 57 9.0 1 3 1 Ditch 110 140 9.0 1 1 2 1 1 Layer 100 100 4.5 1 this fill. They are marsh or water side plants, parsley water-dropwort often occurring near the sea in brackish conditions. The sample from Fill 140 in Ditch 110 resembles that from the upper fill of Pit 49 in that, with the exception of the water crowfoot, the most numerous seeds are those of water plantain, duckweed and rushes, and again fairly shallow water is indicated. A large part of this sample consisted of fragments of leaves, stems and roots. These have not been identified, except that the leaf fragments are probably from broad-leaved trees. All the fragments are small, averaging about 10 mm. in length, with a very small area on either side of the central vein or midrib. In no case is any leaf margin recognisable. Three other fragments can be seen by their vein formation to be monocotyledonous species, but again have not been identified more closely; they might be fragments of grass or reed species. Also in 140 are larval cases of caddis fly, and egg cases of various water fauna and the whole assemblage suggests an undisturbed deposit of shed leaves, seeds and other debris such as would accumulate in a ditch or shallow water course. Taken together the results from the charred and waterlogged groups of seeds illustrate surroundings of damp grassland with ditches and areas of open water. 347 L. BARBER TABLE 4: WATERLOGGED PLANT REMAINS Context Sample vol. (litres) Ranunculus acris/repens/bulbosus Meadow/Creeping/Bulbous buttercups R. sceleratus L. Celery-leaved buttercup R Subg. Batrachium (DC) Water crowfoot (two species) Papaver somniferum L. Opium poppy Papaver sp. Poppy Urtica dioica L. Stinging nettle Chenopodium album L. Fat hen Chenopodium rubrum/glaucum Red/Oak-leaved goosefoot A triplex prostrata/patula Spear-ieaved/Common orache Lychnis flos-cuculi L. Ragged robin Polygonum cfhydropiper (L.) Spach Water pepper P. aviculare agg. Knotgrass Rumex sp. Dock Thlaspi arvense L. Field penny cress Rubus fruticosus agg. Blackberry Epilobium cfmontanum L. Broad-leaved willowherb Hydrocotyle vulgaris L. Marsh pennywort Oenanthe spp. Including: 0. lachenalii Gmelin Parsley water-dropwort and 0. aquatica (L.) Poiret Fine-leaved water-dropwort cf. Apium nodiflorum Fool's watercress Hyoscymamus niger L. Henbane Lamium purpureum L. Red dead-nettle Mentha aquatica L. Water mint Sambucus nigra L. Elder Cirsium cf palustre (L.) Scop Marsh thistle Pit 49 42 9.0 1 c. 500 1 1 1 1 4 57 4.5 4 1 2 3 39 1 Ditch 110 140 9.0 12 2 1 31 29 22 20 2 7 1 4 2 1 1 1 3 9 8 12 348 AN EARLY ROMANO-BRITISH SALT-WORKING SITE AT SCOTNEY COURT TABLE 4: (CONTINUED) C.cfarvense (L.) Scop Creeping thistle Sonchus asper (L.) Hill Prickly sow-thistle Alisma plantago-aquatica L. Water plantain Potamagaton sp. Fennel-leaved pondweed Lemna sp. Common/Ivy-leaved duckweed Juncus spp. - including J.cfeffusus L. Soft rush and J. cf acutiflorus Ehrh. Ex Hoffrn Sharp-flowered rush Eleocharis palustris/uniglumis Spike rush Schoenoplectus lacustris (L.) Palla Common club-rush Carex cfotrubae Podp. False fox-sedge C. cfriparia Curt. Greater pond-sedge Carex sp. (distigmatous) Sedge Bud scales (cf Salix sp.) Leaf, stem and root fragments 103 2 c. 100 c. 50 5 1 2 2 3 2 c.300 11 22 258 4 c. 150 +++ 23 3 2 8 +++ DISCUSSION Unfortunately, too little of the plan of the site remained at Scotney Court to make any real sense of the site's layout and functional areas. This may be in part due to the relatively small area investigated, but is predominantly the result of the majority already having been quarried away. Despite a close inspection of the northern face of the causeway no archaeological deposits were noted here and it therefore seems likely that the focus of activity, including the main salt-working area, had originally been sited to the south and had thus already been destroyed. For this reason, along with the virtual total absence of features which can be shown to be positively connected with salt extraction, such as hearths, it is impossible to compare the site to other excavated salt-working sites elsewhere. The only evidence for salt extraction comes from the presence of numerous pieces of briquetage found in most deposits. Only one feature, the possible evaporation pan noted in the quarry face south of Trench A (Context 73), may be seen as relating directly to salt 349 L. BARBER production, and even this is not certain as too little remained for investigation. Pit 49 in Trench A is also potentially connected with salt production, as it is similar in plan to salt pans found in north Kent at Funton Creek, although the Scotney Court example showed no signs of heating.11 Conversely, the evidence from the seeds in Pit 49 suggests it may have actually been water-filled. If this is the case, it is possibly the remains of an old settling tank which was left to overgrow and was subsequently used for the disposal of domestic rubbish. The briquetage tempering agents, particularly the flint grits, would have been readily available to the site. This, combined with its fragile nature during transport and general crudity strongly suggest briquetage was made on the site when needed. Ancient salt-working methods are described in detail elsewhere and it is thus not proposed to discuss them here, particularly due to the paucity of activity-specific features from the site.12 The environment of the site during the salt production activity in the first century is not fully known. Although the sea now lies some 2 km. to the south, during the period of occupation the site would have been set on marshland close to the sea, probably with a large inlet to the north.13 The exact distance to the sea has to remain unascertained as salt water used for extraction may have reached the site by means of tidal creeks. The environmental evidence indicates, as is to be expected, a predominance of plants which prefer wet, muddy grassland conditions, with expanses of open water. At least one inundation seems to have occurred during the first-century occupation, although the evidence for dating prior to this is scant. This is marked by the shingle layer in Trench D which washed in from the south-west and filled the features open at the time. This did not prevent activity for long, however, as first-century salt-working continued after the deposition of the shingle. Domestic activity is well represented during the first century with the presence of pottery and food remains, notably bone. Shell, however, is obvious by its absence, unless shell middens were sited 11 A. P. Detsicas, 'A Salt-Panning Site at Funton Creek'. Arch. Cant, ci, (1984), 165-8. 12 Colchester Arch. Group 1984. (ibid.) 13 Cunliffe's reconstruction of the coastline during the Roman period puts Scotney Court at the very edge of the coast. B.W. Cunliffe, 1988. Romney Marsh in the Roman Period in (Eds) J. Eddison and C. Green, Romney Marsh Evolution, Occupation, Reclamation. Oxford University Committee for Archaeology, Monograph no. 24, 83-87. 350 AN EARLY ROMANO-BRITISH SALT-WORKING SITE AT SCOTNEY COURT closer to the site's focus. Virtually all the pottery from the site relates to the first -century salt-making activity and is of a utilatarian nature. The majority of pottery is of Kentish types although some may have even been made on site during the manufacture of briquetage. If this is the case, it is more likely to be limited to the cruder grog-tempered vessels. The total absence of samian and other fine wares suggests a low-status occupation with, either no need of finer wares or no way of obtaining them. Fuel for pottery production, and indeed salt-making, is likely to have been a limiting factor for the occupants of the site. Even if the majority of evaporation was done by solar power, hearths would have been needed to finish off the process. As woodland is unlikely to have been present close to the site in any quantity, this commodity must have been brought in, presumably by boat, from drier land to the north and east of the marsh. This is certainly indicated by the presence of wood charcoal from the site which is predominantly of dry soil preference tree species. If local wood was available in any quantities, a higher percentage of damp tolerant species is likely to have been represented at the site. Besides the production of salt, other activities were undoubtedly taking place on the site. These may have included the salting of locally caught fish prior to being transported inland to market as well as elements of animal husbandry. Unfortunately, the bone assemblage from the site is small; however, the overall picture is of a predominance of cattle with a relatively high proportion of sheep/goat. Pig is only represented in small numbers. Only mature specimens of cattle and pig were recovered although juvenile sheep/goat remains are present. Trying to draw conclusions from such a small assemblage is difficult. The limited data available at present may hint at a preference for slaughtering the older animals, perhaps before winter, while the younger ones were taken back to a more permanent base on the highground.14 The younger sheep/goat may have died naturally. It is interesting to note that most bones present do not consist of the 'best cuts' suggesting it is possible that animals were slaughtered before winter and only the selected cuts were taken back to base-camp after being salted. However, logistically, it would be easier to walk the animals back to the base-camp for slaughtering rather than having to carry the meat back after the animals had been butchered. If this is the case the bone on site is likely to represent the off-cuts of animals killed during the site's summer occupation for food. 14 Cunliffe 1988. (ibid.) 351 L. BARBER The plant remains from the site are equally problematic to interpret at present in the light of no comparative material. The presence of charred cereals is interesting as it is unlikely cereals would have been grown anywhere on the marsh at this time. It seems likely that either cereal was brought to the site in small quantities to be processed as needed, to make bread, or the charred cereals are accidentally derived from another source such as chaff, etc., being brought in for briquetage tempering. The general picture one gains of the occupation at the site is not dissimilar to the model put forward by Cunliffe. Although seasonality cannot be proved from the excavated evidence at present, salt-production is only likely to have been possible during the summer months. The presence of animals is well attested and it is likely the drier marsh was used for summer pasture for herds while other activities were undertaken. Fishing seems likely, however, no evidence of use of other marine resources, except salt, was encountered. This is particularly the case for sea-birds and marine molluscs, although further work on the marsh may address this problem. Only one Romano-British feature on the site can be proved to post-date the first century: Ditch 110, Trench D. This is of second to early third-century date. It is possible that Ditch 117 was contemporary as it had a similar stratigraphic position. Whether salt production was still undertaken at this time is uncertain but it seems unlikely; any briquetage in these features, like most of the pottery, is residual from the first-century layers cut through by the ditches. The ditches may well form part of a drainage system associated with a nearby settlement. The presence of the unabraded Colchester beaker (Context 138) would strengthen this suggestion; however, very little pottery of this period is present and it may be the beaker is a piece discarded some distance from the associated settlement. It is possible the Romano-British material located at Lydd Quarry approximately 1 km. to the north-east, which is of a similar date, is the focal point of activity at this time.15 The environment in the second to early third century, as reflected by the drainage ditches, is similar to that of the first century. It is likely the ditches had standing water in them; however, the broad-leaf tree leaf fragments are somewhat more problematic. The majority of wood and charcoal from the site is of trees which prefer dry " I . Greig and M. Gardiner, forthcoming, Excavations at Lydd Quarry. No features of Romano-British date were located, but the quantity of pottery suggested a site very close by. 352 AN EARLY ROMANO-BRITISH SALT-WORKING SITE AT SCOTNEY COURT soils/upland areas and have obviously been brought in from further to the north. The presence of Alnus and Populus/Salix type trees is interesting in that they are more tolerant of wet soils and may have actually grown on the marsh. A number of Alnus twig fragments were found in Ditch 110, fill 140. This, with the presence of the leaf fragments, which are unlikely to have been deliberately brought to the site, suggests the possibility of at least one tree of Alnus type, growing in the close vicinity. No material of third to fourth-century date was located and this fully conforms with Cunliffe's suggestion that the danger of pirate attacks and/or rising sea level caused the marshland settlements, even if only seasonal, to be abandoned. No sign of any sudden termination of occupation was found and it appears the site was abandoned and gradually buried by natural sedimentation. Future work on sites of this period on the Marsh should further address the problems outlined in this report and collect evidence of both domestic and industrial structures. ACKNOWLEDGEMENTS The author is grateful to the Romney Marsh Research Trust for funding the excavation and to Brett's Gravel, through the good offices of Richard Hambley, for the loan of the mechanical excavator and help throughout the on-site works. Thanks must also go to the handful of dedicated volunteers, particularly Jill Eddison and Sue Carrel, as well as the excavation team, Chris Broomfield and Natalie Tompsett, for working in some truly atrocious conditions. Comments and reports from all specialists during the post-excavation stages of the work are also gratefully acknowledged: Sue Hamilton, Nigel MacPherson-Grant, Valery Rigby, Andrew Savage, Wendy Wood, Brian Irving, David Goode, Pat Hinton and Mark Gardiner. REFERENCES Barber, L. A 1996 Roman Salt-working at Scotney Court Archive report, Report no. 134 353