SECTION 7.1: KEMPSFORD STUBBS FARM INTRODUCTIONGEOLOGY, TOPOGRAPHY AND LAND USE EXCAVATION METHODOLOGY Background to the excavationsA programme of archaeological investigation was carried out in an area known as Stubbs Farm, to the east of the village of Kempsford within the parish of the same name in south-eastern Gloucestershire. Oxford Archaeology (OA; then Oxford Archaeological Unit, OAU) carried out the work between June 1991 and July 1995, on behalf of the developer, Multi-Agg Ltd, in advance of gravel extraction. This work included evaluation trenching, watching briefs and limited open area excavations targeted on the two enclosures known from cropmark and evaluation evidence. Further archaeological evaluations were carried out by OAU in areas adjoining to the north (Manor farm, OAU 1992) and west (Multi-Agg Quarry extension, OAU 1998; Fig. 1). Together, these sites revealed part of a Roman landscape incorporating large field systems, multi-ditched enclosures, and a masonry-footed building. Location of site (Fig. 7.1.1: Site location plan)The site lies less than 1 km to the north of the River Thames in the parish of Kempsford, east of the village and on the northern side of the course of the old canal, at grid reference SU 167 970. Geology, topography and land useThe site is situated on the first gravel terrace of the River Thames at a height of around 75 m OD. The ground is generally fairly flat, sloping gently down towards the river to the south, with the exceptions of a slight hill beyond the eastern site boundary. The underlying geology is composed of Oolitic Limestone of Middle Jurassic date. Most of the site is overlain by the gravels of the first terrace, although along the southern part of the eastern boundary there is an area of alluvium and the slight hill to the east is formed of an outcrop of Oxford Clay. The soil in the investigated area was generally silty clay loam. A series of shallow palaeochannels formed a drainage system throughout the area. One group of these features began at the eastern end of the Manor farm site and ran in the direction of Stubbs Farm, almost certainly forming part of the palaeochannel observed running down along the eastern side of the current site, visible in eight of the evaluation trenches. This part of the channel was traceable as a ribbon of alluvium cutting through the first terrace river gravel that underlay the rest of the site. Together these palaeochannels formed a series of abraded streams running along the western side of a slight ridge on the eastern side of Manor Farm and down along the eastern side of the current site to the River Thames. The ridge was also flanked on the eastern side by a similar series of palaeochannels draining to the east. It is unknown when any these streams originated, but they may not have been very ancient. The area occupied by the northern part of former set of palaeochannels was very wet and marshy within living memory. A modern drain down the western side of the long field at Manor Farm and forming the eastern boundary of the Stubbs Farm site is a canalisation of this series of stream courses. The second set has also been canalised into a drain. The fields that made up the site had been ploughed in recent times. The ploughsoil ranged from 0.21-0.25 m deep and directly overlay the natural subsoil. Archaeological backgroundSoil conditions as described above were widespread in the vicinity of the site and ideal for showing cropmarks. Most of what was known of the archaeology in the vicinity, prior to initial investigations in 1991, came from this source or from fieldwalking. Aerial photographs showed linear features crossing the site (Fig. 7.1.1: Site location plan), forming large fields aligned approximately north-south by east-west. In the north-western corner of the site a smaller double ditched rectangular enclosure was visible alongside one of the north-south linear cropmarks (Gloucestershire SMR 3165). North-east of this was a sub-circular enclosure, crossed by both north-south and east-west ditches of the field system (Gloucestershire SMR 3034). Immediately to the north of the site the cropmark system continued at Manor Farm (Gloucestershire SMR 3173 and 3174). There another rectangular enclosure was visible alongside the same north-south linear boundary. This was of a similar size to that on the Stubbs Farm site, but was only single ditched. Evaluation of this site in 1991 dated the field system to the Roman period (OAU 1992). A pair of parallel linear cropmarks, probably a trackway, ran west from the Stubbs Farm site across the adjacent field to intersect with another trackway running NW-SE. Spreads of stone rubble and Roman roof tiles in the area around and to the north of this intersection suggested a masonry building, and this was located during an archaeological evaluation in 1997 (OAU 1998). Some of the north-south aligned cropmarks at Stubbs Farm (Gloucestershire SMR 3155), continue to the south of the site, towards an area of linear and oval cropmarks (Gloucestershire SMR 3166 and 3167). These in turn link into an extensive area of cropmarks, covering approximately 30 hectare north of the River Thames and to the south-east of the site (RCHM Kempsford (7); Leech 1977). To the north and north-east lay the excavated Iron Age and Roman settlements at Whelford Bowmoor (section 6), Thornhill Farm (Jennings et al. 2004) and Claydon Pike (see sections 2-3). The Stubbs Farm cropmarks appeared to form part of the planned agricultural landscape associated with these settlements. Within this system prehistoric elements can be discerned. Part of a large sub-circular enclosure was visible to the west of the Manor Farm rectangular enclosure and was tentatively dated to the Iron Age. Another complex of cropmarks, which includes circular ditched enclosures, lying 1.5 km to the north-east of Stubbs Farm, has been dated by fieldwalking to the Iron Age. Excavation methodologyArchaeological investigations began in June 1991 at Stubbs Farm, Kempsford, Gloucestershire, after a programme of mineral extraction, and subsequent ground restoration, was proposed by Multi-Agg Ltd. Gravel was to be extracted from an area of 19.08 hectares, in an area where cropmark evidence suggested the presence of archaeological remains. In response to conditions placed on the planning consent to record archaeological remains on site in advance of the work, OAU was commissioned to investigate the archaeology of the area. Investigation of the site was carried out in a number of phases over the next four years. Field Evaluation (Fig. 7.1.2: Field evaluation)The initial two phases of the investigation in June and September 1991 took the form of field evaluations. The first of these involved excavation of five trenches in the south-western part of the site, which was the first part to be stripped in preparation for gravel extraction. Trench 1 of this first phase took the form of a salvage area excavation, while the other four trenches were standard 50m by 1.8m evaluation trenches. A further 35 similar trenches were excavated across the rest of the site in the second phase of evaluation, adding up to a one percent sample of the site. In each trench the topsoil was stripped by machine to reveal any archaeological features cut into the natural gravel. Any features located in this manner were then sampled by hand to determine their nature and depth and to recover dating evidence. The larger ditches were excavated by machine. Excavated features were recorded in plan and section. The machine stripped ploughsoil and machine excavated ditch fill was monitored to recover finds. This work revealed a regular Roman field system incorporating a rectangular enclosure as suggested by the cropmark evidence. The field system was shown to cut across a circular enclosure, but the character and relationships of the other features were not fully established. A further phase of work was required in order to do this. This was to involve two complementary elements: strict archaeological monitoring of the stripping of the whole site and area excavation around both the rectangular and circular enclosures. Watching briefThe planning condition for the gravel extraction specified a watching brief to be carried out over the entire area. The topsoil was to be stripped over areas rather than having a working face. This was done using a 360° excavator with a toothless bucket, subject to archaeological supervision. Excavation (Fig. 7.1.3: Plan of excavated site)An area of approximately 60 x 90m around the rectangular enclosure and 60 x 50m around the circular enclosure was machine stripped under archaeological supervision. All discrete features were hand excavated by a minimum volume of 50%. Approximately 10% by volume of the ditches was hand excavated to determine their character, form, and stratigraphic sequence and to recover datable artefacts. When the circular enclosure could not be clearly dated using this strategy a different strategy was employedied. The entire volume of the ditches was machine excavated and the spoil hand sorted to recover datable artefacts. Only 14 sherds of pottery were recovered in this way. The surface of the site was examined by Mark Maillard using a metal detector, and several metal objects (SF 1027, 1023, 1024 and 1025) and half a dozen pieces of lead were located by this means. Unfortunately, vandals removed many of the latter before they were archaeologically recorded. Significantly, nothing was recovered from the surface of the field prior to the excavation by local detectorists. Chronology and phasingDue to extensive damage by burrowing animals, the preservation of features and stratigraphy on the site was poor, and most of the ditches had spatial rather than stratigraphic relationships. This disturbance may also have contaminated some deposits with later pottery, although nearly all recovered dating evidence suggested that activity was largely confined to the 2nd century AD. An attempt has been made in this report (section 7.3) to describe the elements of the site in a chronological order, but because of the limited remaining stratigraphy, this could not be adhered to rigidly and many features remain undated. |