Predicting archaeo-colluvium on the Berkshire Downs
A new and relatively unexplored source of archaeological and environmental information on the Chalkland are the sediments of dry valleys. From relatively recent studies these deposits have been shown to be the product of ancient land use and soil loss and a rich repository of both primary and archaeological material. There have, however, been no attempts to determine the spatial distribution of this class of ancient valley deposit - for the purpose of this project termed archaeo-colluvium. The study focuses on the north-west part of the Berkshire Downs (58km2), a landscape of both chalk and superficial drift deposits, which together with relict ancient fields, lynchets and settlement features offered a suitable area in which to develop and test a method for predicting the distribution of these deposits. The project uses a multi-disciplinary approach which combines the traces of ancient arable (Celtic fields), some topographic parameters and assumptions about past soils, erosion regimes, and colluvial preservation in the development a GIS-based predictive model for the distribution of archaeo-colluvium within this study region. Following the production of a map of predicted sites, archaeo-colluvium was checked in the field by an extensive auger and trench survey of the main valleys and tributaries. Dry valley sequences were assembled and dated, both for the purposes of evaluating the accuracy of the model and to draw some archaeological implications from the nature and distribution of these deposits. A synthesis of local soils, colluvium and wider archaeological observations suggested a three-stage chronology for prehistoric and Romano-British land use on the Berkshire Downs, centred around earliest arable use at Seven Barrows with later shifts to surrounding catchments of mixed chalk and clay soils. The results of the field program were encouraging with the predictive model verified at 71% of target sites. A number of field observations were contrary to the original model, notable among these were that thicker sequences were common in landscapes of heavier soils as was evacuation of valley sediments from open chalk landscapes. This feedback allowed some modified principles to be briefly tested on three other Chalkland landscapes.