Evidence of Late Roman Settlement near the site of the Church Hall, Kemsing

Evidence of Late Roman Settlement near the site of the Church Hall, Kemsing

sean wallis

with contributions by

Luke Barber, Ceri Falys, Felicity Howell and Malcolm Lyne

An archaeological excavation in advance of the construction of a new church hall at Kemsing revealed traces of a late Roman settlement dating from the fourth century ad. The finds suggest that the settlement may have been of relatively high status and it is possible that a masonry building stood in the vicinity, though no below-ground remains of this were located.

An archaeological excavation was carried out in July 2019 by Thames Valley Archaeological Services, immediately to the west of Kemsing Parish Church (TQ 5556 5878) (Figs 1 and 2), on behalf of the Parochial Church Council. Planning permission had been granted by Sevenoaks District Council for the construction of a replacement church hall, subject to a condition which required the implementation of a programme of archaeological work. An evaluation in October 2011 (Wallis 2011) had demonstrated the site’s potential and showed it to contain Roman features which might be damaged or destroyed by the development. As a result, excavation was required to mitigate these effects and to enhance understanding of the nature of these features.

The site is located immediately west of St Mary’s Church, and north of the historic part of Kemsing village. It is relatively flat, at a height of 99-103m aod. The underlying geology consists of Lower Chalk.

Archaeological background

The site lies within an area of moderate archaeological potential derived from its location close to the historic core of Kemsing and its parish church. In general, the setting of the village (on a spring line) is one which was typically attractive for prehistoric, Roman and Saxon occupation. A Roman building, apparently of at least four rooms with hypocaust heating, was excavated in 1949 at Dynes Road, but published only in very summary form (Anon 1950: no. 8 on Fig. 1), and a single coin of Constantine was found in Montford Road (Pyke 1982).

The village is first mentioned in a document from ad 822, as Cymesing, or ‘place of a man called Cymesa’ (Mills 1993). The historic village is centred on a well dedicated to St Edith, the illegitimate daughter of King Edgar, who was supposedly born in Kemsing in ad 961. The parish church stands to the north of the village centre. A church was probably built here in the late Saxon period, although the present nave dates from the twelfth century. It has been suggested that the seventeenth-century Grade II Listed Building to the west, known as The Keep, overlies an earlier motte and bailey castle, although direct evidence of this has so far proved elusive.

An evaluation of the site was carried out in 2011, when there were plans to build a new vicarage to the west of the existing church hall. Two trenches were excavated, one either side of the hall, and late Roman features were uncovered. Trenches were also excavated close to the existing vicarage, to the south-east of the church, but just one post-medieval pit was found in that area. The Roman features recorded in the evaluation are included in the site description below.

the excavation

Based on the results of the evaluation, the excavation was designed to record any archaeological deposits within the footprint of the new church hall and to address research questions essentially aimed at elucidating the nature and extent of any prehistoric, Roman or Saxon activity on the site.

The excavation area of approximately 244m² was stripped down to the top of the underlying natural chalk, which generally necessitated the removal of up to 0.50m of topsoil (50) and subsoil (51) deposits. The ground reduction was slightly less deep in the area previously under the former church hall, and there had been some disturbance from the footings of this building and its associated service runs. A small area of 4m² was later stripped in advance of a new soakaway and subject to a watching brief.

All of the archaeological features in the excavation area, including post-holes, gullies and pits, were sampled by hand (Figs 3 and 4). Following half-sectioning and recording (as shown on Fig. 3), the majority of the discrete features were fully excavated in the hope of obtaining more finds. The relatively shallow depth of the cut features suggests that the area had been ploughed in the distant past, and this is supported by the fact that the features were not clearly visible within the subsoil horizon. Two late Roman pits were recorded in the northern half of the excavation area along with three linear features of similar date. In the southern half of the excavation there were three post-medieval pits, along with five post-holes which yielded no dating evidence. An undated tree throw (22) was recorded in the soakaway area. Despite a programme of sieving of bulk soil samples, no charred plant remains were recovered and only tiny amounts of small charcoal fragments, all unidentifiable. The post-medieval features are shown on plan and described in the archive but not further discussed here.

The archaeological features containing late Roman pottery (c.ad 325-400) were all in the northern part of the excavation area, although it is considered possible that the undated post-holes to the south may also date from this period. All feature fills were mid to dark brownish grey clayey silt or silty clay, unless otherwise noted.

Linear features: Ditch 1000 was aligned approximately east to west across the northern part of the excavation area. Its terminus (1) had been recorded in evaluation trench 2, to the east of the excavation but it continued beyond the excavation area to the west. The ditch was generally between 0.62m and 0.82m wide, and up to 0.20m deep. A slot (8) was dug by hand through the eastern section of the ditch, whilst another slot (20 /21) investigated the relationship between Ditch 1000 and the gully which extended north from it. Ditch 1000 had a single fill (52/60/80) which produced nine sherds of late Roman pottery, along with numerous fragments of animal bone and a small piece of burnt clay.

The relationship between Ditch 1000 and Gully 1001 could not be established as their fills were too similar, and it is likely that the two features were contemporary anyway. The gully was up to 0.62m wide and 0.20m deep, and contained four sherds of late Roman pottery and a small fragment of animal bone (from deposit 76). It extended north towards Gully 7, but the area where the two features presumably met had been disturbed by a modern service.

Gully 7 was aligned approximately east to west across the northern end of the excavation area, roughly parallel to Ditch 1000. Its western part had also been destroyed by the modern service trench. The gully was up to 0.34m wide and 0.14m deep, with near vertical sides and a flattish base. The only finds from its fill (59) were fragments of animal bone and oyster shell.

In evaluation trench 1, to the west of the excavation area (Fig. 2), two gullies (2 and 3) were aligned north to south, with slightly irregular edges, roughly 0.75m apart. Gully 3 was 0.15m deep and up to 0.68m wide, with a flattish base and a single fill (54) from which three small sherds of late Roman pottery were recovered. Gully 2 was more substantial perhaps deserving to be called a ditch, 0.8m wide and 0.4m deep, with steep sides and a flat base. Its single fill (53) contained animal bone, along with thirty-four sherds of late Roman pottery, including 6 sherds from one dish which may be fifth-century, two very small fragments of tile, weighing 21g, and a small number of fragments of fuel ash slag and fired clay. It truncated the western side of a small pit (6). The two features probably marked a single boundary, perhaps with a hedge between them.

Discrete features: Two sub-circular pits (17 and 19) were investigated in the northern part of the excavation area, either side of east-west Ditch 1000. The largest pit (17) measured 2.56m by 1.90m, and was up to 0.70m deep with relatively steep sides, a flattish base, and several distinct filling deposits (70-4 and 82) were visible in section (Fig. 4). Deposit 82 consisted of a light brownish grey marly chalk, and was only recorded close to the pit’s edges. It probably derived from the pit edges collapsing slightly shortly after the feature was originally dug and, unsurprisingly, contained no finds. The first true fill of the pit was deposit 74, which consisted of dark greyish brown silty clay with occasional chalk inclusions. Six sherds of late Roman pottery were recovered from this deposit, along with a large number of oyster shells, and two fragments of animal bone. The layer immediately above (73) consisted of mid brownish grey silty clay, with moderate amounts of flint and chalk. No finds were recovered from deposit 73, but the layer of dark grey silty clay (72) above it yielded two sherds of late Roman pottery, along with fragments of tile including possibly tegula, and animal bone and two oyster shells. A relatively thin layer (71), largely consisting of small chalk fragments, was recorded immediately above deposit 72. Deposit 71 produced no finds, and could represent deliberate backfilling of the pit. The uppermost layer of the pit (70) was a dark brownish grey silty clay which contained two sherds of late Roman pottery, along with a residual prehistoric sherd. The deposit also contained over one hundred fragments of animal bone, along with a few pieces of tile, including tegula, daub and weathered stone. The presence of the tile, daub and stone suggests that it may be derived from the demolition of a nearby building. One of the tile fragments from deposit 70 appeared to be post-medieval peg tile, and intrusive.

Pit 19 had been truncated by a modern service trench, but it originally measured at least 1.90m by 1.50m, and was up to 0.40m deep. No finds were recovered from its primary fill of chalky silt (78), but the layer above (79) contained eleven small sherds of late Roman pottery and a few fragments of animal bone. Further fragments of bone were found in the uppermost fill (77), along with a piece of Roman tegula.

In evaluation trench 1, Pit 6 had originally been 0.52m long and at least 0.18m wide. It contained several fragments of animal bone and a small sherd of late Roman pottery. A similar pit (5) was only partially exposed within the trench, but was 0.13m deep. Animal bone fragments were recovered from its fill (57), along with a single sherd of late Roman pottery.

Undated features: Five probable post-holes were recorded in the southern part of the excavation area. Three of these (12, 13 and 14) formed a line just to the north of Pit 15, whilst the other two (10 and 11) were roughly parallel about 5.30m to the south. If, as suggested, the area had been disturbed by ploughing in the past, there may have been further post-holes present originally. All were roughly 0.35-0.50m in diameter, though all slightly off-circular, and mainly under 0.20m deep. No finds were recovered from the post-holes, although their fills were generally more similar to those in the Roman features to the north, than those in the nearby post-medieval pits. It is therefore possible that the post-holes may represent the remains of a small late Roman building, situated to the south of the small enclosures repres-ented by Ditch/Gullies 7, 1000 and 1001.

finds

Pottery by Malcolm Lyne and Luke Barber

The evaluation yielded 42 sherds (393g) of pottery from six contexts and the excavation added 37 sherds, weighing 220g, from nine contexts (Table 1). These totals include three post-medieval sherds, excluded from the table, and one prehistoric sherd. Overall the pottery consists of small to medium-sized sherds with slight to extensive signs of abrasion. As such at least some of the material appears to have seen notable reworking though the majority has not. The single tiny prehistoric sherd in Pit 17 (70) is very abraded and clearly residual.

The bulk of the assemblage is of the Roman period and mainly of Late Roman date wherever diagnostic sherds are present. Although rim sherds are completely absent from the excavation assemblage, the wares themselves are quite characteristic of the third and fourth centuries. The less diagnostic unsourced sandy wares could be slightly earlier but there is no reason they need be. Considering the small size of the assemblage there is quite a high proportion of finewares suggesting the associated household was of some standing.

All of the assemblages were quantified by numbers of sherds and their weights per fabric. These fabrics were identified using a x8 magnification lens with inbuilt metric graticule in order to determine the natures, forms, frequencies and sizes of added filler inclusions and were, for the most part, classified using the fabric codes formulated by the Canterbury Archaeological Trust (Macpherson-Grant et al. 1995). One exception is the possible sub-Roman fabric SR1, where no such coding previously existed. None of the assemblages are large enough for any further quantification by Estimated Vessel Equivalents (EVEs) based on rim sherds.

By far the largest and most significant pottery assemblage is that from Gully/Ditch 2 in evaluation trench 2 (context 53). These 34 sherds include:

eight abraded fragments in Late Roman grog-tempered ware fabric LR1.1 (c.270-420),

one from a jar in Harrold Shell-tempered fabric LR3 in similar condition (c.350-420),

four sherds from C49 and C51 bowls in Oxfordshire Red Colour-coat fabric LR10 (c.240-420),

one from an M22 mortarium in Oxfordshire Whiteware fabric LR22

(c.300-420),

three fresh sherds in Alice Holt Greyware fabric LR5 from a type 6A.8 dish (c.330–420) (Lyne and Jeffries 1979) and a jar,

three from a rilled jar in Overwey/Portchester D fabric LR6 (c.330-420).

A fragment from a closed form in Much Hadham Oxidized ware LR13 (c.250-420) and seven in miscellaneous unattributable greywares are also present, as are six fresh sherds from a shallow, carbon-soaked, handmade dish in a silty fabric SR1 with sparse chopped grass filler. The fragments from this dish are the freshest in the assemblage and in a fabric not dissimilar to Early Saxon ones with similar filler of c.450-650 date. The form does, however, have more in common with Late Roman straight-sided dishes than the few recorded, somewhat deeper, open forms of Early Saxon date. A late fourth- to fifth-century date can be inferred for this important assemblage, which appears to have continued to accumulate into the sub-Roman period.

The other pottery assemblages are much smaller and include a flake in Overwey/Portchester D fabric LR6 (c.330-420) from the fill of Pit 6 in Trench 6, two fragments from an everted-rim jar in Late Roman grog-tempered ware and one from a necked jar of uncertain form in Alice Holt greyware from Gully 3 in Trench 3 and one fragment in Late Roman grog-tempered ware from Pit 5 in Trench 5.

Roman fabrics

R7. Miscellaneous greywares

LR1.1. Late Roman Grog-tempered ware

LR3. Harrold Shell-tempered ware

LR5. Alice Holt greyware

LR6 Overwey/Portchester D fabric

LR10 Oxfordshire Red Colour-coat

LR13 Much Hadham Oxidized ware

LR22 Oxfordshire Whiteware

?Sub-Roman fabric

SR1. Handmade carbon-soaked fabric with profuse silt and sparse chopped grass inclusions.

Ceramic Building Material by Luke Barber

A relatively small assemblage of brick and tile was recovered, in mixed condition, with the Roman types being more abraded and the post-medieval ones quite fresh. The assemblage is detailed in archive. The ceramic building material assemblage is mainly composed of Roman tile. The pieces are generally small and somewhat undiagnostic of form, though most identifiable fragments appear to be tegula.

Animal Bone by Ceri Falys and Felicity Howell

A total of 286 pieces of animal bone, weighing 3786g, was recovered from ten late Roman features in the excavation and evaluation combined. Overall preservation was generally good, although many fragments displayed occasional root etching of the cortical bone surface and a moderate to high degree of fragmentation was present in most contexts, limiting identification.

Initial analysis roughly sorted elements into general size categories: large (horse or cow), medium (sheep/goat, pig or deer), and small (e.g. dog, cat, etc.). Where possible, a more specific identification to species of origin was attempted. The minimum number of individuals (MNI) was determined based on duplication of skeletal elements and/or differences in the stage of skeletal/dental development.

Information regarding the MNI was primarily derived from the remains recovered from pit 17, with a minimum of two cattle, three pigs, one sheep/goat and one deer, and at least one ‘small’ animal, of unidentifiable species. However, 44 per cent of the bones present could not be sorted even into a general size category, due to their small size and undiagnostic appearances.

The highest proportion of identifiable pieces of bone were allocated to the ‘large’ size category (36.8 per cent). Cattle were identified in Ditch 1000 (slots 8 and 20), in addition to Pit 17. A minimum of two cattle was suggested by the presence of two mandibular fragments in Pit 17 (fill 70), which were of notably different sizes and states of dental eruption. A portion of cow calcaneus from Pit 17 (72) displayed evidence of butchery, as it had been bisected down the longitudinal midline by a sharp implement.

A total of 44 pieces of ‘medium’-sized animal bone were recovered from five features: pits 5, 6, 17 and 19, Gully 1001 (slots 1 and 18) and Gully 2. Again, most of the identifiable elements were from Pit 17 (70), which suggested an MNI of five: three pigs, one sheep/goat and one deer. The identification of three pig individuals (two juvenile and one older animal) was made based on the presence of maxillary and mandibular fragments which displayed varying degrees of dental eruption and tooth wear. A sheep/goat was indicated by the recovery of a maxillary fragment with in situ teeth, and a left distal humerus, while a deer was represented by a small portion of antler. Beyond Pit 17, more pig teeth were identified in Pit 5, while a pig mandible with in situ teeth and a canine fragment were identified from Ditch 1000, slot 1 (52). Fragments of pig left ulna, radius and humerus, as well as two right mandibular canines were found in Gully 2 (53). Sheep/goat was also represented by a molar fragment from Pit 6 (58), and two maxillary molar fragments, one mandibular molar and one right metatarsal from Gully 2 (53).

Evidence for a minimum of one, unidentified ‘small’-sized animal was also recovered from Pit 17. A total of eight small pieces of long bone and rib shafts were present in deposits 70 and 72.

Finally, a small assemblage of oyster (Ostrea edulis) shell, weighing just over 200g, was found in Ditch 1000, Gully 7 and Pit 17.

Discussion

The excavation at Kemsing Church Hall revealed a modest number of archaeological features, the presence of which had been indicated in an evaluation of the site in 2011. The three linear features in the northern part of the excavation area may represent small enclosures, possibly related to stock control. These produced a small assemblage of late Roman pottery, largely dating to the fourth century, as did the two probable refuse pits situated nearby. The larger of these pits appears to have been left open for quite some time as several distinct layers were clearly visible in its backfill. The uppermost deposit contained fragments of tile, fired clay and weathered stone which may represent demolition rubble from a fairly high-status Roman building. The presence of such a building in the vicinity is also suggested by the high percentage of finewares amongst the small pottery assemblage, and a small piece of Roman glass recovered from post-medieval Pit 15 to the south. Two lines of post-holes were recorded to the south of the Roman linear features but none yielded any dating evidence. However, their fills were similar to those of the Roman features, so it is possible that they represent the badly truncated remains of an ancillary building within the Roman settlement. The focus of this fourth-century settlement may have been a masonry building which could have stood anywhere around the present site. Traces of this building may survive in the vicinity of the church hall, although if it was to the north, east or south it is likely to have been destroyed by the cemeteries.

The presence of a Roman settlement in this area should not come as a surprise given the local topography. The River Darent is a tributary of the Thames and would have been a much larger waterway in the past. It rises near Westerham and breaks through the North Downs at Otford, flowing northwards towards the Thames. It is joined by the River Cray before entering the Thames between Crayford and Dartford Marshes, where it is tidal. During the Roman period it was necessary to establish a ford where the road from London to Dover (Watling Street) crossed the Darent, and there is growing evidence for an associated roadside settlement at the aptly named Dartford. It is likely that a similar settlement would have existed at Otford, where the Pilgrims’ Way (despite its later name, an ancient North Downs trackway) crosses the River Darent, and this is supported by the significant evidence for Roman activity recorded in the area. The Darent Valley has one of the highest densities of villas in Britain, with traces of buildings recorded at Dartford, Wilmington, Darenth, Farningham, Horton Kirby, Longfield and Ash. Closer to the present site (Fig. 1), the remains of villas or other substantial Roman buildings have been Eynsford (Lullingstone) [Fig. 1: 1], Shoreham [2], Otford [3-7] and Kemsing itself [8], albeit details on the latter are scant (Anon 1950). Further evidence of Roman activity has been found at several sites close to the river [9-13]. A large cemetery dating from the first to fourth centuries, and containing over 100 cremation burials, was excavated at Otford [14], and further burials from the Roman period have been discovered in the area [5, 15-16]. Recent archaeological excavations to the south of Kemsing have revealed Roman features, including a small enclosure [17], and two further square enclosures have been identified from aerial photographs [18-19].

Kemsing itself may have been quite an important site in the past due to the fact that a spring (St Edith’s Well) rises in the area now occupied by the village. The water from this spring feeds into a small tributary of the River Darent. Although this watercourse now resembles little more than a drainage ditch, it must have been a more prominent feature in the past, as it acted as the parish boundary between Kemsing and Seal. It is also worth considering the importance of the Pilgrims’ Way, the section of an ancient trackway which took pilgrims from Winchester to the shrine of St Thomas Becket in Canterbury. The trackway runs, in one form or another, all the way from Wiltshire to Folkestone, for much of its length along the southern slopes of the North Downs. It has certainly been in use since prehistoric times, and the cluster of Roman sites around Otford suggests that it was a significant route during that period.

acknowledgements

The work was commissioned by Mr Roger Molyneux of Molyneux Architects, on behalf of St Mary’s, Kemsing, Parochial Church Council. The investigation was carried out to a specification approved by Mrs Wendy Rogers, the Kent County Council Archaeological Officer, who advises Sevenoaks District Council. The fieldwork was undertaken between 9th and 18th July 2019, with a further watching brief in May 2020, supervised by the author, assisted by Will Attard, Virginia Fuentes, Daniel Haddad and Tom Stewart, with Felicity Howell also assisting in the evaluation. The site code is MCK11/80 and the archive is presently held at TVAS South, Brighton, and will be suitably deposited in due course.

references

Anon., 1950, ‘Report for 1949 [Kemsing: Excavations on Roman building carried out by Messrs C. H. Brett and L. Chandler]’, Archaeologia Cantiana, 63, xliv-v.

Lyne, M.A.B. and Jefferies, R.S., 1979, The Alice Holt/Farnham Roman Pottery Industry, CBA Res. Rep., 30, London.

Macpherson-Grant, N., Savage, A., Cotter, J., Davey, M. and Riddler, I., 1995, Canterbury Ceramics 2. The Processing and Study of Excavated Pottery, Canterbury.

Mills, A.D., 1998, Dictionary of English Place-Names, Oxford.

Pyke, J.A., 1982, ‘Sevenoaks Branch’, Archaeologia Cantiana, 98, 260.

Wallis, S., 2011, ‘Kemsing Vicarage, Kemsing, Kent: an archaeological evaluation’, unpubl. TVAS rep. 11/80, Brighton.

Fig. 1 Site Location, showing known Roman sites in the area around Kemsing (their numbers referred to in text).

Fig. 2 Plan showing excavated area in relation to evaluation trenches.

Fig. 3 Plan of all excavated features.

Fig. 4 Feature sections.

Table 1. Catalogue of Pottery

Cut	Fill	Fabric	Form/note	No.	Weight (g)
1	52	R7	Jars	2	17
2	53	R7	Closed form	7	38
2	53	LR1.1	Jars	8	45
2	53	LR3	Jar	1	4
2	53	LR5	Jar; convex-sided dish	3	35
2	53	LR6	Rilled jar	3	16
2	53	LR10	C49 bowl; C51 bowl	4	32
2	53	LR13	Closed form	1	4
2	53	LR22	M22 mortarium	1	34
2	53	SR1	Dish (sub-Roman?)	6	144
3	54	LR1.1	Everted rim jar	2	7
3	54	LR5	Everted rim jar	1	7
5	57	LR1.1	Jar	1	4
6	58	LR6	Jar	1	2
8	60	LR1.1		3	24
8	60	NFRS	Flagon	1	8
16	69	R7		1	2
17	70	LR1.1		2	16
17	70	flint gritted	prehistoric	1	1
17	72	LR1.1		1	2
17	72	Sandy blackware		1	8
17	72	R7		1	2
17	74	LR1.1		3	14
17	74	LR6		1	14
17	74	LR10		2	2
18	76	LR1.1		2	3
18	76	MR7		1	2
18	76	NVCC		1	18
19	79	R7	refired	1	10
19	79	Sandy-shelly ware		10	12
20	80	LR1.1	bitone	1	2
20	80	LR10		1	2
20	80	LR5	combed arc decoration	1	28