In February 2003 Archenfield Archaeology conducted an archaeological evaluation of the former Brooklyn Ford Car dealership, Bath Road, Worcester on behalf of Neil Grinnall Homes and Spa Housing Association. A geoarchaeological study of cores taken from a borehole situated close to the line of Birmingham to Worcester Canal showed that the deposits consisted of an accumulation of alluvium overlying bedrock and overlain by a greater depth of made-ground. The lower levels were shown to have formed in waterlogged conditions, and samples may contain valuable pollen and plant remains that relate to the former course of the Frog Brook. Archaeological trial trenching showed that the site has been very disturbed by 19^th and 20^th century construction and demolition. Deep accumulations of made ground and a build up of layers of industrial dumping and the presence of back filled cellars were recorded. No evidence of occupation observed on either side of Bath Road was earlier than the later post medieval period.

The geoarchaeological survey

By David Jordan, Terra Nova

Summary

Deposits studied at a site on Bath Road, Diglis, Worcester were found to consist of an accumulation of alluvium overlying bedrock and overlain by a greater depth of made-ground. The lower alluvium was probably deposited gradually in stable, wet conditions and may preserve pollen and other evidence by which the environmental history of the valley floor might be reconstructed. It is unlikely, however, that the sequence of environmental development could be dated since the deposits do not contain material on which dating could be reliably carried out.

Aims of the study

This study aimed to clarify the origins of deposits recovered from a site at Bath Road, Diglis, Worcester and assess their potential to provide geoarchaeological and palaeoenvironmental evidence in response to a brief issued by Worcester City Museum Archaeology Section (Brief 02/37).

Background

The deposits of the Frog Brook valley, in which the site lies, may be of archaeological and palaeoenvironmental significance both in themselves and in the evidence that they might preserve. The valley lies close to the centre of the historic town and, in accumulating sediments from its catchment and from the river Severn, may have also accumulated human occupation debris representing urban development.

We do not have accurate maps which show the location of the site in relation to the topography of the valley before it was greatly altered by 19^th century development. The geological 1:50000 map (BGS, 1993) and modern topography, however, suggest that the site sits near the middle of the former valley and may overlie deep alluvium. The geological map shows alluvial deposits overlying rocks of the Triassic Eldersfield Mudstone Formation.

The Frog Brook is thought to have developed as a Late Devensian river channel. It has been used as part of the city defences and, later, as the town ditch. The valley is known to have been prone to flooding as water from the River Severn ponded headwards during floods from pinchpoints downstream at Callow End and Cliffey Wood (Morris, 1974). Thus the Frog Brook basin deposits might record the changing flood regime of the river Severn at Worcester through the Holocene as well as that of the Frog Brook itself.

These backwater alluvia have the potential to preserve excellent sedimentary evidence in an environment dominated by slow water movement and minimal erosion. This also means that there is the potential for the survival of well contexted archaeological and environmental evidence in the alluvium and former ground surfaces of the Frog Brook valley, through waterlogging and accumulation.  

Previous studies of the Frog Brook alluvia (Terra Nova 2000, 2002) have not encountered deep deposits and the potential for a sedimentary record incorporating archaeological evidence is yet to be realised.

The natural soils

The natural soils of the site have not been mapped but those of nearby valleys, similar to that of the Frog Brook, are Pelo-alluvial gleys of the Compton series. These are mottled soils of moderately good drainage which are periodically waterlogged by a fluctuating ground water table (SSEW, 1986).

Method

The site had undergone a geotechnical borehole and test-pit investigation, before our own fieldwork, the results of which suggested the presence of deep artificial deposits (made ground) which posed problems of investigation and of site safety. The test pits and bores suggested that the north and eastern side of the site did not overlie Holocene alluvium. A bore at the north-east corner (WS2) identified sand and gravel at 0.6m which might have been redeposited material overlying alluvium but, from pits and borehole records nearby, appeared suggest more likely to be a natural, Devensian terrace deposit. Test pit SG2 on the north-west corner of the site identified blue grey sandy clay at 1.7 m which may have been natural alluvium as may that identified at 3.9m by pit SG3. Bore WS1 identified alluvium at 3.7m and, from SPT data it is inferred that the Mudstone bedrock lay at about 5 m.

We were significantly hampered, in analysing the geotechnical data, by the lack of absolute heights OD for the tops of the pits and bores which made it difficult to relate the records to each other and to the surrounding area. The varying deposit records and depths of made ground may indicate, however, that some soil (potentially including alluvium) may have been removed from the site and, subsequently, replaced with mixed make-up layers during the development of the area. It seemed possible, therefore, that the site did not contain any continuous alluvial deposits and that any, apparently natural, sequence of deposits should be interpreted with care.

We originally planned an attempt to recover samples from any buried alluvium using a narrow percussion window corer in two locations in the south western part of the site where the existing records suggest that we have the greatest chance of success.

We found, however, that the deep excavations that this would have required were made impossible by the dangerously loose, deep made ground. Deposits being recovered by a much larger diameter shell-and-auger percussion borehole in the south-western corner of the site showed that a single, deep sequence of intact deposits might be recovered in this way using U100 core sample tubes. The core being carried out at the time recovered good evidence that there was, indeed, a sequence of alluvial deposits beneath the made-ground.

We therefore changed our plans and carried out a further shell-and-auger core which recovered a complete sequence of intact samples in 10cm diameter plastic tubes from 2.5 to bedrock at 7.4 metres below the ground surface. 

We observed the excavation of trial pits to the east of Bath Road. These showed that the bedrock there is much shallower and that mudstone, probably disturbed by periglacial mixing and solifluction, lies at only 2 metres from the modern ground surfaces and shallows rapidly uphill to the east. The rock and periglacial deposits are here overlain by a deep, natural soil profile which appeared to be formed in natural colluvium. Above and cut into this was a sequence of apparently natural archaeological deposits and more recent dumps of debris. The buried soil and artificial deposits above showed signs of prolonged waterlogging due to impeded drainage and the rise of groundwater in the valley bottom. This was much less pronounced, however, than in the deposits from the boreholes to the west, closer to the centre of the valley and the course of the former Brook itself.

Observations

The deposits recovered as intact U100 tube samples were examined in the laboratory. They consisted of a sequence of mostly clayey deposits from the base of the industrial-debris made ground to the late Devensian gravel deposits at about 7 metres. It was striking that this sequence differed considerably from that which we originally observed only 1 metre to the west.

We observed nine well-defined strata, which are described in the Appendix using criteria adapted from Avery (1980) and Jones et al. (1999). The sequence broadly consists of :

1      Bedrock, with a disturbed upper surface, overlain by 0.75m of gravel and sandy deposits. The gravel and sand were moderately well sorted and appear to have been deposited from fast-moving water, most probably under late Devensian conditions

2      0.75m of mid-grey clay, sandy at the base and increasingly tinged blue towards the top. There were abundant fine former root pores within this deposit, concentrated towards the upper surface, and filled with darker and apparently more organic fine matter within a reduced matrix. There was, however, no trace of the former ground surface from which the roots and the dark, fine matter had descended.

3      1.2 metres of clayey deposits, a strong red-brown and slightly stony at the base becoming mid-grey towards the top. This deposit also had abundant former root pores descending from above its upper surface, containing a reduced blue-grey clay.

4      2.0 metres of mid-grey clay containing small fragments of ceramics becoming larger and more abundant towards the surface. This deposit also contained a few small fragments of wood as well as ash and charcoal which became dominant, mixed with red-brown clay, at the surface.

Above this was a further 2.5 metres of dark, ashy, stony made-ground to the modern surface.

These four zones of deposit appear to have formed in quite separate phases of deposition and represent different periods of change and stability in the Frog Brook valley.

Deposit 1 represents the decay and mixing of the bedrock surface under Devensian periglacial conditions, followed by alluviation from energetic stream-flow, most probably in the late Devensian or very early Holocene.

The clayey deposit 2 above appears to be entirely natural and represents gradual alluvial accretion in a low-energy, slow-flowing stream environment during the Holocene. The upper surface of this deposit has been truncated but it is very difficult to estimate how much has been lost. No surface, from which soil development has taken place, survives and the remaining deposits appear to represent only the lower part of a soil profile into which deep roots have penetrated. Thus at least 50cm or so is likely to be missing. It is possible, however, that even more has been removed since the deposit is likely to have accumulated gradually over a long period of time and the root pores found within it may have formed from surfaces which were subsequently buried by further accumulation. In such deposit former periods of stability may be represented by an increase in root density but further alluviation may still have added a considerable depth of deposit above.

The lower part of deposit 3 consists of largely unaltered material derived from the bedrock surface and is unlikely to have been deposited naturally, unless by a slump from the slope above – though in this case we might expect to find some more organic soil matter mixed into the deposit, which is not the case. The deposit is very slightly sorted, suggesting that it was deposited in moving water, and the increasingly grey clay above probably represents gradual accumulation of further stream alluvium and the reworking of the material beneath over a long period. There must have been sufficient periods of stability on the valley floor for further rooting to develop and, therefore, for a soil profile to form. This profile, however, appears also to have been truncated before the overlying deposits accumulated.

Deposit 4 is unstructured and consists of a mixture of clay and occupation debris which appears to have been deposited in a few, brief episodes. There is little evidence of subsequent reworking and it appears more likely that the deposit became sealed by further, overlying deposits which prevented the development of a soil profile.

The made-ground above is of entirely artificial origin and consist largely of industrial debris unaltered by further weathering.

Magnetic susceptibility measurements made on the cores support these conclusions. Deposits 1 and 2 and the lower part of deposit 3 show only low values with a maximum of 10 SI at the top of deposit 2 – a normal value for a natural material with some soil formation given the mineralogy of the deposit parent material. The upper part of deposit 3 and the whole of deposit 4 show higher values rising to more than 200 SI at the top of deposit 4. This is what we would expect of deposits derived in part from industrial and occupation debris and also shows the gradual increase in the proportion of debris which makes up the deposits towards the surface as well as the abrupt change in mineralogy and deposit source at each successive boundary.

Discussion and conclusions

Of these deposits only 1 and 2 are likely to have accumulated naturally and to contain palaeoenvironmental evidence which may be used to create a reliable understanding of the development of the valley floor. All the strata above are probably derived from elsewhere and deposited by human activity. They do contain evidence for some gradual accumulation and reworking under natural conditions but this will have taken place within derived material and it will probably not be possible to separate that environmental evidence which is genuinely informative from that which is derived and thus misleading.

The lower deposits, 1 and 2, can give us some useful palaeoenvironmental information. We note, in particular, that deposit 2 represents gradual alluvial accumulation under waterlogged conditions and that pollen and other evidence may survive within it. The mineral matter of the deposit, however, does not preserve any useful evidence of the sedimentary regime since the deposit accumulated gradually and has been heavily reworked by roots and soil fauna. Thus individual episodes of flooding and the development of the river regime is not represented stratigraphically. The deposits appear to have remained waterlogged for almost all of the time since they accumulated since they show no sign of soil structure formation, drying cracks or the movement of sesquioxides to form mottles.

To sum up – the Mudstone bedrock is weathered and appears to have been broken-up and disturbed, most probably under Devensian periglacial conditions. Gravel and then sand were deposited from flowing water in the later Devensian and the stream flow gradually reduced as vegetation stabilised in the early Holocene. It is likely that the stream rationalised into a single, stable channel during the Mesolithic and that the site had a wet soil subject to regular inundation from the river Severn and from the Frog Brook itself. We found no peat but it is possible that peat beds developed under very wet conditions nearby. We cannot say for certain when the overlying layers were deposited on the site although we can guess at the activity which they represent. Historic maps suggest that the site was only developed in the post-Medieval period and it is therefore likely that deposit 2 represents alluvium which accumulated continuously from the beginning of the Holocene to this time. We lack, however, any material from deposits 1 and 2 which is suitable for radiocarbon or other forms of physical dating and it may therefore be impossible to assess when deposition occurred.

Further evaluation of pollen and diatoms is likely to be worthwhile but no reliable plant macroscopic remains, mollusc or insect remains were noted in the deposit despite careful preparation.