HISTORY OF MEDIEVAL POTTERY RESEARCH IN THE REGION

With the exception of purely descriptive reports, such as that on the collection of pottery from the Pithay, Bristol (Pritchard, 1926), medieval pottery studies in the region began in the 1930's. Notable early reports are those on Lydney Castle and a pit group from Old Sarum, associated with a coin of William the Conqueror (Casey, 1931; Stone and Charlton, 1935).

From then on, and especially in the 1940's and 1950's, E. M. Jope and G. C. Dunning studied much of the pottery included here, together with material not now available for study, and produced a series of distribution maps for medieval pottery types in Southern England (for example those included as appendices to the Selsley Common report, Dunning 1949, and the Ascot Doilly report, Jope 1959). Interpretation of these maps was limited because it was not then possible to characterise the pottery by fabric analysis and therefore a plotted type might be the product of one or several centres. In fact, there was no way of telling whether a distribution resulted from the movement of pottery or from the transmission of ideas. Another problem was that their maps did not indicate how common a type of pottery was at the marked findspot although Jope did use a symbol on his maps to show 'negative evidence', sites which ought to have produced that type if it were in common use. The culmination of Jopes work on pottery distribution was an essay in Culture and Environment (1963), where medieval pottery style distributions were compared with those found in church architecture, place-names and other artefacts to show the amount of regional variation in material culture. These are some of the few studies of their time in medieval archaeology which use archaeological data as a source of information independant of medieval history.

Barton's study of Ham Green ware was published in the same year (Barton, 1963). This paper, together with a study of the medieval jugs of Worcester published by the same author, marked the beginning of detailed ware studies in the area (Barton, 1967b). The Ham Green report contained a statistical table of the typological characteristics found on wasters from the kiln site and these, plus the visual appearance of the pottery, were used to plot the distribution of the ware. Dating was based on the stratigraphic association of Ham Green pottery at Bristol and Cheddar.

In the late 1960's and early 1970's there was a dramatic increase in the number of medieval sites excavated as well as in their size and the complexity of stratigraphy discovered. Gloucester Museum appointed a Field Officer in 1968 (H. R. Hurst) whilst P. A. Rahtz initiated a series of excavations on the defenses of Hereford in the same year (Victoria Street, the Brewery Site, see Shoesmith, 1982). Excavations on Bristol Castle by the Field Officer of Bristol City Museum also began in 1968. Perhaps the largest exacavation in the region has been on the Gloucester Telephone Exchange Site (Site 77/69). This site produced an extremely complicated sequence of buildings dating from the 10th/11th centuries to the late 13th/14th centuries with isolated features of later medieval date. Excavation continued into the late 1970's on a reduced scale but, writing in 1983, it now appears that this phase in the development of medieval archaeology is over and that much of the potential of pottery studies for medieval archaeology in the region will not be realised.

Further developments in the potential of the study of medieval pottery in the region came as a result of the work of D. P. S. Peacock on Neolithic, Iron Age and Romano-British pottery (1969, 1968 and 1967a). Peacock demonstrated that undecorated, coarse pottery was not always produced in the immediate locality of the site (itself an unexpected conclusion) and that it could be characterised without recourse to typology. Henceforth it was possible to study distribution and typology separately, and to make much more sense of both. Once petrological groupings had been made Peacock found that there were indeed typological differences as well, although there was less than perfect agreement between the two. It was therefore possible to use typological characteristics in certain cases as an aid to source identification. The same phenomenon has been noted and used in this study.

In the early 1970's Hodder published a series of studies of Romano-British pottery distribution using various methods of quantification to illustrate the rate at which frequency declined with distance from the source (Hodder, 1974a, 1974b). This work suggested that three factors influenced the distribution of Romano-British pottery, namely, the road network, the use of waterways and the influence of the marketing system. The distribution of Rowland's Castle ware jars was skewed to the north. Hodder showed that this was probably due to the use of a road for transport of pottery. The rate of fall-off in frequency of Oxfordshire wares could be divided into two; on sites with access to the Oxfordshire kilns by water, the fall-off was markedly less steep than on sites not situated near waterways. The distribution of Savernake Ware showed that the fall-off in frequency was not regular around the kilns but was instead centred on the nearby town of Mildenhall. These studies showed that in the Romano-British period considerable information about the mechanics of local trade could be obtained through a study of the coarseware pottery industries.

More recent developments have included the use of textural analysis to characterise sands (the methods are described in Shackley, 1975), the application of Heavy Mineral Analysis, notably Williams' study of Black Burnished Ware (1977), and the application of Neutron Activation Analysis to characterise Stamford Ware and medieval floor tiles (Killmurry, 1980, and Hughes et al., 1982). These methods would enable petrologically 'bland' quartz-sand tempered wares and untempered 'fine' wares to be characterised. Unfortunately none of these methods has the 'feedback' of the petrological approach, which enables inclusions to be reliably identified by eye once their identity has been verified in a sample using thin-section analysis.

The research climate has altered beyond recognition from the early 1960's, when the lack of suitable analytical tools produced an impasse in interpretation and little stratigraphic data existed, to the 1980's when the archaeologist is faced with a plethora of characterisation methods and techniques of distribution analysis and a large body of excavated data. The methods chosen for this study, and the reasons for their choice, are described below.