SECTION 7.3: KEMPSFORD STUBBS FARM FINDS REPORTS

7.3.1 ROMAN POTTERY (Paul Booth)
Introduction
Methodology
Fabrics
Vessel types
Chronology
Illustrated Vessels
General discussion
Bibliography

7.3.2 COINS (Paul Booth)

7.3.3 THE SMALL FINDS (Hilary Cool)

7.3.4 THE CERAMIC BUILDING MATERIAL (Kate Atherton)
Introduction and methodology
Tile types
Conclusions

7.3.1 ROMAN POTTERY by Paul Booth

Introduction

The excavations produced some 907 sherds of pottery weighing 9120 g. The material was all of the Roman period apart from a single flint tempered sherd which may be assignable to the Iron Age. The bulk of the pottery was probably of 2nd century date. The sherds were generally in moderate to poor condition. Much of the material was quite badly fragmented, though recent breaks were discounted as far as was possible, and surfaces were often heavily eroded, making identification of fabrics difficult in some cases and removing much evidence for decoration. The erosion may be in part a consequence of soil conditions rather than post-depositional processes on the site. An additional characteristic of the soil, noted in connection with other pottery assemblages from this area of the Upper Thames valley, is that it imparts a reddish brown tinge to some sherds. This also hampers the ready distinction between oxidised and reduced wares.

Top of Page

Methodology

The material was examined by context and sherds assigned to fabric/ware categories within the Oxford Archaeological Unit's Roman pottery recording system. Quantification was by sherd count, weight and EVEs. Vessel types were recorded, along with details of rim and base forms and decoration where appropriate. The full records are on sheets which are contained in the project archive.

Fabrics

Identification of fabric was at a fairly generalised level, usually at an intermediate stage of the fabric/ware definition hierarchy used in the OAU recording system. The major ware groups represented in the Kempsford assemblage were: S - samian ware, F - fine wares, A - amphora fabrics, M - mortarium fabrics, W - white wares, Q - white-slipped wares, O - oxidised `coarse' wares, R - reduced `coarse' wares and B - black-burnished ware. Most sherds were assigned to subgroups of these categories (eg R30, a general grouping for moderately fine sandy reduced wares), though some were identified at the level of specific fabric (eg M22, Oxfordshire white ware mortaria). In view of the problems of identification arising from the physical characteristics of the sherds (see above) and the fairly small size of the assemblage, detailed recording of the fabric of each sherd did not seem to be justified.

Brief descriptions of the fabrics present in the group, or familiar names of well-known wares, are given below. Fuller descriptions can be found in the documentation of the recording system contained in the project archive. Each description is followed by the total quantities of the fabric.

S20. South Gaulish samian ware. 1 sherd, 2 gm.

S30. Central Gaulish samian ware. 14 sherds, 223 gm.

F60. Fine oxidised fabric with red brown colour-coat. Source uncertain. 1 sherd, 4 gm.

A11. South Spanish amphora fabric, as used for Dressel 20 etc. 5 sherds, 84 gm.

M21. Verulamium region white mortarium. 3 sherds, 187 gm.

M22. Oxfordshire white mortarium. 3 sherds, 73 gm.

M31. Oxfordshire white-slipped mortarium? 1 sherd, 6 gm.

W11. Oxfordshire parchment ware. 1 sherd, 115 gm.

Q10. Fine sandy oxidised fabric with ?white slip. Source unknown. 1 sherd, 6 gm.

O10. Fine oxidised `coarse' ware. 8 sherds, 69 gm.

O20. Coarse sandy oxidised ware. 2 sherds, 2 gm.

O30. Moderately fine abundantly sandy oxidised ware. 151 sherds, 737 gm.

O51. Oxidised fabric with moderate-common rounded clay/grog pellets. 4 sherds, 71 gm.

O80. Miscellaneous coarse (usually ?grog-tempered) oxidised fabrics. 7 sherds, 140 gm.

R10. Fine reduced `coarse' ware. 1 sherd, 2 gm.

R20. Coarse sandy reduced ware. 5 sherds, 70 gm.

R30. General moderately fine reduced fabrics. 41 sherds, 403 gm.

R35. Moderately fine abundantly sandy reduced ware. 218 sherds, 1864 gm.

R38. Grog and sand-tempered reduced fabric. 43 sherds, 1671 gm.

R41. Slightly sandy reduced ware with grog and organic inclusions and prominent mica. 4 sherds, 38 gm.

R95. Coarse grog-tempered reduced/oxidised fabric. ?Savernake ware. 13 sherds, 531 gm.

B10. Possible black-burnished ware, source uncertain. 24 sherds, 154 gm.

B11. Black-burnished ware, probably Dorset BB1. 355 sherds, 2661 gm.

FN4. Dark grey-black fabric with abundant well-sorted angular flint inclusions. 1 sherd, 6 gm.

The two principal components of the assemblage were reduced coarse wares and black-burnished ware, with oxidised wares of lesser importance and the combined `fine and specialist' wares (samian, fine wares, amphorae, mortaria, white and white-slipped wares) totalling only 3.3% of sherds (7.7% of weight). There were substantial differences in the comparative representation of the principal ware groups in terms of sherd count and weight. Both oxidised wares and black-burnished ware were rather better represented in sherd count (respectively 19% and 41.8% as opposed to 11.2% and 30.9% of weight) while the reverse was true of reduced wares (35.8% of sherds, 50.2% of weight). This discrepancy is a result of two main factors, firstly the high degree of fragmentation of black-burnished and (particularly) oxidised wares (the average weight of sherds in the principal oxidised fabric, O30, was only 4.9 gm) and secondly, the converse situation with some of the reduced fabrics, in that R38 and R95, which occurred as sherds of greatly above-average weight, were sufficiently numerous to affect the average weight of the combined reduced coarse ware group. It may be noted that the principal reduced ware fabric, R35, had a similar average sherd weight (8.6 gm) to that of black-burnished ware fabric B11 (7.5 gm). The latter was the most common individual fabric in the assemblage (by sherd count and weight, but not by EVEs).

As would be expected, the coarse ware fabrics for the most part indicate domination of the assemblage by local production sources. The principal oxidised and reduced ware fabric groups, O30 and R35, have the fine abundant sand tempering characteristic of the North Wiltshire industries, the nearest production centre of which was at Purton, only some 13 km SW of Kempsford. Sources of oxidised fabrics other than O30 are uncertain. O10 may have been an Oxfordshire product, and the occurrence of a tankard in O51 could indicate either a Severn Valley or North Wiltshire source for the fabric, as the form was characteristic of both industries. Fabrics R38 and R95, like R35, are probably North Wiltshire products. R95 may be Savernake ware, but is basically grog-tempered without some of the other inclusions, such as flint, usually found in this fabric. The distinction between R95 and R38, also principally grog-tempered, was relatively slight and one of degree rather than character. R38 was probably from a source closer to Kempsford than the Savernake Forest and may perhaps also have been from the Purton area. The only other fabric in this group, R30, is unsourced, but could include products of the Oxford industry. The source of the very distinctive micaceous fabric R41 is unknown, but may also have been within the Upper Thames region.

The principal non-local fabric was B11. It is possible that this fabric is slightly over-represented in the present figures because of the problems of identification mentioned earlier. Since the surfaces characteristic of this fabric were usually eroded away there was a tendency to consider any black coarse sand-tempered sherds as potentially black-burnished ware, while in fact there may have been a degree of overlap with fabric R35. In particular, the majority of sherds assigned to B10, from a single vessel with peculiarities of decoration (an external BB2-like wavy line and internal lattice) which indicate that it was not a Dorset product, may possibly have been a version of R35. Despite these problems there is no doubt that Dorset BB1 formed a significant part of the assemblage. The similarity between B11 and some probable sherds of R35 raises the possibility that there were local attempts to copy black-burnished ware, however. In some cases it could be seen that the similarity extended not only to characteristics of the fabric, but also to its finish and decoration and the range of forms, particularly the occurrence of jars of `cooking pot' type.

The `fine and specialist' wares require little comment. The small samian assemblage, almost entirely of Central Gaulish origin, included no decorated sherds. Fine wares were notably absent, though the erosion of surfaces may mean that a few oxidised sherds which were originally slipped escaped detection. The number of such cases is likely to have been very small, however. The single identified fine ware was a rim of a small beaker of 2nd century type but of uncertain source, though it is just possible that this sherd was of fabric F65, originating in the Upper Thames region and with a 2nd-4th century date range. Mortaria were from the Oxford region and (almost certainly) from the Verulamium region, though the sherds in this fabric were very worn and the grits distinctive of this source were therefore absent. The sole white ware sherd was a large part of an Oxford parchment ware bowl of the common type P24.

Top of Page

Vessel types

As with the fabrics, the identification of vessel types was usually at the generalised level of major classes (jar, bowl etc) or their principal subdivisions (such as `cooking pot type' jar or carinated bowl), except where specific well-established types (eg examples of Young's (1977) Oxfordshire typology) were present. Quantification of vessel types was by rim EVEs. Since the overall EVEs total of the assemblage (8.58) was relatively low the true importance of some of the lesser types may be difficult to judge, but the relative proportions of the principal types reflect a plausible picture. The correlation of vessel types by fabric is given in Table 1.

The assemblage was dominated by jars (vessel class C), which constituted over 69% of the total EVEs, with uncertain jar or bowl types (class D) a further 2.8% of the assemblage. Jars occurred in oxidised, reduced and black-burnished fabrics. While totalling just over half of EVEs in oxidised fabrics, however, they amounted to c 89% and 79% respectively of vessels in reduced and black-burnished wares. No other vessel classes were particularly important. Straight sided dishes (class JA) were the second most common type, 7% of the total EVEs, while bowls, which in most assemblages in this region would be expected to be second only to jars in importance, amounted only to 3.9% (classes HA and HB together), though uncertain bowl/dish forms (class I, where insufficient of the profile survives to determine the height:diameter ratio crucial to determining the precise form of the vessel) were a further 2.2% of the assemblage. Bowls and dishes, with the exception of the carinated bowl (Young P24) in Oxford parchment ware, were mostly in black-burnished ware and the reduced fabric R35, though the only rim fragments of samian ware were in the indeterminate bowl/dish category. A single ring necked flagon rim (class BA) in fabric O30 constituted 6.1% of the total EVEs and beakers (class E) amounted to 4.2%. These types, and the only tankard represented, occurred mostly or entirely in oxidised fabrics. Specialist forms were generally not represented by rims. The sole mortarium rim was of Young type M3, of 2nd century date (Young 1977, 68-70).

Top of Page

Chronology

A number of aspects of the character of the assemblage suggest that it is largely of 2nd century date. The ubiquitous grog-tempered and related `Belgic type' fabrics so characteristic of the late Iron Age-early Roman period in the region are completely absent from this assemblage. The only possible pre-Roman sherd is the single fragment in the flint-tempered fabric FN4. The regularity of the inclusions in this fabric might suggest a date in the late Iron Age rather than earlier, but this is by no means certain. This isolated sherd apart, the absence of `unromanised' fabrics precludes a start date for the assemblage before the Flavian period at the very earliest. As indicated above, the principal fabrics in the assemblage are probably North Wiltshire products and Dorset black-burnished ware. The North Wiltshire kiln sites in the Swindon area, including that at Purton, are thought to have developed in the Hadrianic period (Anderson 1979, 9). Similarly, the large-scale advent of black-burnished in the region is unlikely to be earlier than about AD 120, the `traditional' date for the expansion of this industry (Gillam 1976, 57), though some pieces may have arrived before this date. The Kempsford black-burnished ware, as far as can be seen, occurs entirely as common 2nd century (and occasionally later) forms, and there is no reason to believe that any of this material is significantly earlier than c AD 120.

Black-burnished ware was recorded in 39 of the 74 contexts which contained pottery. While some of these identifications, particularly of single small sherds, may be open to question (see above), there is no doubt that the fabric was found in 14 of the 18 context groups which contained more than 10 sherds. Black-burnished ware was therefore an important element in the assemblage throughout the site, and from the point of view of chronology it is irrelevant whether the Dorset original or more local imitations were in use, since the latter are most unlikely to have preceded the former. It may therefore be that occupation on the site did not start significantly before c AD 120 at the earliest.

As already indicated, the majority of the black-burnished ware encountered was of 2nd century types. There were a few exceptions to this, including a single bowl with a crude bead and flange rim unlikely to date before the later 3rd century. There were also a few examples of burnished lattice decoration (on cooking pots) of a style suggesting a date of at least the mid 3rd century (ie tending to obtuse rather than acute angled lattice), but these were greatly outnumbered by examples of acute angled lattice decoration.

The bulk of the datable pieces of the most common fabric in the assemblage were therefore of the 2nd century. The local reduced and oxidised ware repertoire is less closely datable, though these industries may have been in decline in the later 2nd century (Anderson 1979, 9). Negative evidence, such as the absence of characteristic late Oxfordshire products (with the sole exception of the single example of type P24) and other typical later Roman fabrics and forms, also suggest that activity of later 3rd-4th century date was on a much reduced scale. There were only five context groups which on ceramic criteria were dated after the end of the 2nd century, and only two of these (1179 and 1198) are likely to have been late 3rd century or later. It is of course possible that some of the less well-dated reduced and oxidised wares postdated the 2nd century, but there is no certain evidence for this at present.

The few examples of fine and specialist ware types all suggest, or are consistent with, a 2nd century date for the bulk of the assemblage. This is also indicated by the balance of the vessel types. The relatively high representation of jars, at about 69% of the total vessels, is broadly comparable to figures for the early/mid 2nd to 3rd century phases in Area A at Asthall (between 62.4% and 71.2%, unpublished), whereas earlier and later representations of jars at the same site are respectively higher and lower, figures consistent with a generally observable (though not often quantified) trend. Cumulatively the evidence suggests that the most intensive activity on the site was probably in the period from the early/mid 2nd century to the late 2nd/early 3rd century, and that it continued thereafter at a much reduced level at least up to the end of the 3rd century, if not a little later. The ceramic evidence is insufficient to indicate if this later activity was continuous.

Top of Page

Illustrated Vessels (Figs 7.3.1a-c)

A limited number of representative vessels were selected for illustration. All were from ditch fills. They are presented as context groups where possible. In each entry the details of the vessel are followed by the context information.

Figure 7.3.1a: Pottery 1

1. Fabric R38, type CD, medium mouthed jar. 1603, ?top fill of middle ditch of circular enclosure.

2. Fabric F60, type E, beaker. 1055, fill of inner ditch of rectilinear enclosure.

3. Fabric R35, type C, jar. 1112, fill as above.

4. Fabric M22, type K, mortarium (Young 1977 type M3). 1047, fill as above.

5. Fabric O30, type BA, narrow ring-necked flagon. 1047 as above.

6. Fabric R30, type CD, medium mouthed jar. 1047 as above.

7. Fabric B11, type CK, `cooking pot type' jar. 1047 as above.

Figure 7.3.1b: Pottery 2

8. Fabric B11, type CK, `cooking pot type' jar. 1047 as above.

9. Fabric B11, type CK, `cooking pot type' jar. 1047 as above.

10. Fabric R30, type CD, medium mouthed jar. 1098, fill of outer ditch of rectilinear enclosure.

11. Fabric O51, type GA, tankard. 1095, fill as above.

12. Fabric R30, type CD, medium mouthed jar. 1095 as above.

Figure 7.3.1bc: Pottery 3

13. Fabric B10, type JA, straight sided dish. 1095 as above.

14. Fabric B11, type JA, straight sided dish. 1320, fill of N-S ditch on W side of rectilinear enclosure.

15. Fabric W11, type HA, carinated bowl (Young 1977 type P24). 1198, fill as above.

16. Fabric B11, type JA, straight sided dish. 1376, fill of gully to S of rectilinear enclosure.

Top of Page

General discussion

In terms of the fabrics and sources represented the assemblage appears typical of sites in the region, though because of its' small size and relatively restricted date span the range of fabrics is relatively restricted. This may also have been a function of socio-economic status. A survey of assemblages from the region (Booth forthcoming) indicates that low status sites (expressed in archaeological terms, i.e. of site layout, structures and artefactual evidence) of 1st and 2nd century date have fine and specialist ware representations below 5% (of sherd numbers), and that contemporary sites with a higher level of fine and specialist ware are villas or related rural sites and roadside settlements and towns. The Kempsford assemblage falls squarely in the low status bracket, with a fine and specialist ware representation of 3.9% of sherd count, contrasting eg with figures of 7.5% for the villa at Roughground Farm (Green and Booth 1993, 141) and 7% for Asthall. Nearby Thornhill Farm had only 0.7% fine and specialist wares, but this very low figure is only characteristic of sites with intensive 1st century activity, many of which terminate in the early-mid 2nd century (cf Lambrick 1992, 82). The position of Kempsford within the low status group is emphasised, however, by the fact that the base level of fine and specialist wares increases significantly in the later Roman period and any site, even of low status, occupied through the 3rd century would be expected to have a fine and specialist ware level above 5%. The fact that this is not the case at Kempsford seems to confirm its relatively low status.

The most distinctive features of the Kempsford assemblage relate to its chronological range. As mentioned above, a number of rural sites in the region originate in the late Iron Age (in some cases earlier) and terminate in the 2nd century. Local examples include Thornhill Farm and Neigh Bridge, Somerford Keynes, though a late phase of activity on the latter site is indicated by metalwork. A few of these sites, such as Claydon Pike, continue in use throughout the Roman period. Otherwise, sites occupied through the late Roman period often appear to have their origins in the late 1st-early 2nd century, as at Rough Ground Farm. The peculiarity of the Kempsford assemblage is that it does not continue to the end of the Roman period. In this respect it is comparable to an assemblage of c 50 kg of pottery from Whelford Bowmoor, only c 2-3 km distant, which appears to be of very similar character and also to date largely to the 2nd-3rd centuries. A further characteristic which the two assemblages have in common is a high representation of black-burnished ware (this comment is based on a subjective view of the Whelford assemblage, which has not been quantified). This may be a local peculiarity. Black-burnished ware amounted to 21.5% of all sherds from the 1990 excavation at Roughground Farm (Green and Booth 1993, 135) and to c 23.5% of sherds at Claydon Pike (unpublished). These figures, while quite high, are considerably less than the representation at Kempsford. At Asthall Area A black-burnished ware does not rise above 10% of the phase sherd totals until the 3rd century, when it reached 17.5%, so here, despite a major road location, there was considerably less black-burnished ware than at Kempsford. Further down the Thames Valley black-burnished ware, while always present on rural sites, was not particularly common. At Yarnton, for example, it only amounted to less than 2% of the total sherds, and c 5.6% of the 3rd-4th century pottery from the site. While there is extensive evidence for a surge in the distribution of black-burnished ware at the end of the 3rd century AD (M Lyne pers comm) this dominance of markets in the 2nd century appears to be a more localised phenomenon. It may be related to a marketing pattern based on Cirencester, but in the absence of quantified data from excavations there this cannot be demonstrated. Alternatively, the possibility that there was some production of black-burnished ware within the region may need to be reconsidered.

Top of Page

Bibliography

Anderson, A S, 1979, The Roman pottery industry in North Wiltshire, Swindon Archaeol Soc Rep 2

Booth, P, forthcoming, Quantifying status, some pottery data from the Upper Thames Valley, J Roman Pottery Stud

Gillam, J P, 1976, Coarse fumed ware in North Britain and beyond, Glasgow Archaeol J 4, 57-80

Green, L S, and Booth, P, 1993, The Roman pottery, in Allen, T.G., Darvill, T.C., Green, L.S. and Jones, M.U. Roughground Farm, Lechlade, Glos.: a prehistoric and Roman landscape, Thames Valley Landscapes: The Cotswold Water Park 1, (OUCA) Oxford

Lambrick, G, 1992, The development of prehistoric and Roman farming on the Thames gravels, in M Fulford and E Nichols (eds) Developing landscapes of lowland Britain, the archaeology of the British gravels: a review, Soc Antiqs London Occ Papers 14, 78-105

Young, C J, 1977, Oxfordshire Roman pottery, Brit Archaeol Rep (British Series) 43

Top of Page

7.3.3 COINS by Paul Booth

Of the 25 small finds two were coins. One (SF 1023) was found in the evaluation backfill of the large N-S field ditch 1275 and probably came from the fills of the ditch. It is a sestertius with a bust on the obverse and the legend IM]P TRAIANO AUG[. The reverse has a standing figure (1) with it right hand outstretched; the legend in invisible.

The second coin (SF no 1013) found by metal detectorist Mark Maillard came from a spot adjacent to the domestic dump in the N side of 1274. It was a dupondius [?] of 24 mm diameter. There are traces of a radiate crown on the reverse with the legend LIBERTAS C[ and a bearded head on the obverse. This may be an dupondius of Antonius Pius (138-161).

Both of these cons are in poor condition with corrosion and damage etc. but neither was very heavily worn when lost. This suggests that they were lost within the 2nd century.

Top of Page

 

7.3.3 THE SMALL FINDS by Hilary Cool

All of the stratified finds were found in the fills of ditches 1273 and 1274 of the double ditched rectangular enclosure. If the obviously modern and the undiagnostic material from the topsoil is excluded from consideration, there are 17 items that that could relate to the Roman occupation. Nine of these are iron nails (quantified by numbers of heads) and two are undiagnostic iron fragments.

Of the remaining material the only relatively closely dateable item is the foot of a T-shaped brooch (1025) found unstratified (Fig. 7.3.2: Brooch). The hollow back and forward facing foot knob cell identify it as coming from a Hull Type 132 or Nor'Nour brooch (Hull 1967, Type 17). This example is more elaborate than the type normally is but side mouldings as here are occasionally noted (eg ibid 38 no. 85, fig. 16). The form belongs to the Lower Severn tradition of T-shaped brooches, though the majority of examples still come from the site of Nor'Nour on the Scilly Isles. It is not closely dated but a later 1st to 2nd century date is appropriate.

The other items consist of a lead pottery repair clamp (1019) retaining a fragment of reduced pottery and three iron shoe cleats (1010 and an example without small find number from context 1204).

Top of Page

 

7.3.4 THE CERAMIC BUILDING MATERIAL by Kate Atherton

Introduction and methodology

The excavation produced an assemblage of 45 pieces of Roman ceramic building material with a total weight of 2332g (Table 1). Each fragment was weighed and its fabric assessed. All surviving dimensions were measured and any distinctive features were noted. All fragments were assigned to one of several types: tegulae roof tile, imbrices roof tile, tubuli or box-flue tile and flat tile with a thickness less than 39 mm. No brick fragments (tiles with a thickness greater than 39 mm) were present. Fragments with no surviving dimensions and no recognisable features were weighed and classified as miscellaneous fragments.

Tile types

Six fragments of tegulae (total weight 571 g) were identified by the presence of a flange or, if the flange itself was missing, by a groove that characteristically appears at the base of the flange. No signature marks, tally marks or prints were present. On assessment, two fabrics appeared to be present. Four of the fragments were made from a white/pink soapy fabric (Fabric 3) and two were manufactured from a uniform, smooth, orange fabric (Fabric 2). One of the latter had a flange surviving which had a total height of 60 mm and the tile was 27 mm thick. The other complete flange (SF 1009) was made from the other fabric (Fabric 3) and this had a height of 50 mm and a thickness of 20 mm. Three other small fragments of flange were found that were made from the same fabric.

Only one fragment (72 mm) of imbrex was found although it is possible that there are other examples among the flat tile and miscellaneous fragments that are too small for the profile to be apparent. This fragment was made from the same white/pink fabric as the majority of the tegulae, suggesting that both forms of roof tile were from the same source. The thickness of the tile was 16 mm which is typical for tiles of this type.

Three pieces of tubuli or box-flue tile were found (765 g), although one (SF 1117) was in a fragmentary condition. The pieces were worn and abraded and no combing was visible. Two pieces showed traces of burning. One (context 1032) appeared to be made from a similar fabric to the tegulae and imbrex (Fabric 3) but was slightly more orange. The other fragments were made from Fabric 1 which was a slightly coarse, mixed clay matrix with grog inclusions.

There are four pieces of flat tile (288 g), two of which (contexts 1060 and 1110) are made from fabric 2 and have a thickness between 30 and 33 mm and are, therefore, probably tegulae fragments. The remaining two (contexts 1047 and 1096) are made from a soapy orange fabric that has been noticeably better fired than the other tile and has therefore been assigned another fabric type (Fabric 4) although it is possible that differences in firing technique may be the reason for variances between this fabric and Fabric 2. These two fragments have a thickness between 14 and 17 mm and are therefore possibly imbrex or, as is more likely, tubuli fragments.

The remaining 31 pieces (636 g) have no surviving dimensions and can only be classified as miscellaneous fragments. The majority of the fragments (26) are made from Fabric 1 and may, therefore, be further fragments of tubuli. This is supported by the fact that there is evidence of burning on several of the pieces. Four fragments (contexts 1056 and 1060) are made from fabric 2 and one (context 1110) is made from Fabric 3.

Top of Page

Conclusions

The assemblage is not a large one but few fabrics are represented and it is likely that there was only one, local, source for the tile fragments. The fabrics are all similar with the main difference being one of colour. The majority of the fragments are worn and there are relatively few surviving dimensions. This is partly because of the soft nature of the fabrics but it may also be partly due to subsequent disturbance of the tile. The presence of tubuli, as well as the more usual tegulae and imbrices, suggests a substantial building of some status in the vicinity of the site.

The Oxford Archaeological Unit evaluation at Kempsford Roman ‘Villa’ in 1998 uncovered a sample of Roman building material that was notably made from a similar fabric to the ones represented at Kempsford Stubbs Farm (Atherton 1998). It is, therefore, a reasonable hypothesis that the tile from Stubbs Farm came from the same source as the tile from Kempsford Roman ‘Villa’, if not from the ‘villa’ itself.

Top of page