CHARACTERISATION.

Wares were characterised by a combination of thin-section analysis, examination of sherds with a x20 binocular microscope and a certain amount of study of the local clay sources. An attempt was made to study at least one thin-section of every fabric (see Appendix 1) and in the case of wares distributed over long distances (ie. over 40 miles from their supposed source) thin-sections of samples from both the centre and extremities of the distribution were examined.

Textural sub-divisions of fabrics were not made unless the pottery naturally split into such groups (a subjective assessment was made in these cases). Therefore the fabric groups used here probably vary more in texture (grain-size and quantity of inclusions) and in colour than is normal practice in medieval pottery reports (Robinson, 1979).

Conversely, the identity of the inclusions was considered of paramount importance. For example, the pottery of the Malvern Chase encompasses a wide range of textures and colours but it all contains igneous rock fragments and is treated as only four distinct fabrics. A sample of 40 thin-sections was analysed by point-counting and the results are plotted as a ternary diagram (Vince, 1977a, fig.1). This shows that the relative proportions of quartz silt to quartz sand to Malvernian rock can be used to define the fabrics. With practice this is possible visually.

The only wares which could not be characterised by the presence or absence of inclusion types, or by their relative proportions, were those containing solely quartz sand temper or those with practically no inclusions. In the study region these wares were not as ubiquitous as they appear to be elsewhere and their exclusion is not too serious. At both Hereford and Gloucester exclusively sand-tempered wares formed less than 10% of all pottery whilst in London, for instance, the proportion is closer to 90%.

The use of the binocular microscope was invaluable. No useful results would have emerged without it, whilst with its use thin-sectioning was often redundant, revealing no further information. The binocular microscope in fact revealed some types of data lost in thin-section, for example the surface features of quartz grains and details of iron ore inclusions, which are opaque in thin-section. The binocular microscope is also much quicker to use than thin-sectioning and this is an important consideration when analysing pottery in bulk. The ideal sequence of analysis would be as follows: the first site to be examined (of its date) in an area would be processed slowly, examining a fresh edge of each sherd to identify and describe the inclusions. On the basis of this a type fabric series would be prepared and examined in thin-section. This would result in a modified type fabric series which would be used to classify the original pottery, and any other collections of similar date from the same region. Eventually it should be possible to identify the majority of fabrics by eye, using the binocular microscope on groups of wares where it is known that visual analysis alone is misleading. This is approximately the proceedure followed at Gloucester, Hereford, Bath, Chepstow and Newbury and consequently the results from these towns are the most reliable in the region. Elsewhere modifications were made through lack of equipment or restrictions on time.

Clay analysis was carried out on a random basis, except in the Malvern Chase where an attempt was made to define the potential clay sources for the wares by sampling clays throughout the parish of Hanley Castle. All of these clays were essentially Keuper Marl which had been weathered and mixed with overlying sands and gravels or had been re-deposited in stream banks. It was revealing that many of the samples when fired were not similar to any Malvern Chase wares known. Only one good match was found, between samples from Hanley Swan and the Malvern Chase 'pink' fabric (see Ch.2). The amount of local variation in the clay sources in their weathered and contaminated state was very high and the composition of clays between the sampled points could not therefore be extrapolated. However, when a good match was found it was likely to be a reliable indication that the clay source used by the potters was close by.

This should mean that pin-pointing of the clay sources for known kilns is possible although other areas showed much less variation in their clay sources (see Ch.2). For example, clays from the Lower Lias in the Severn Valley and the Cotswolds contained virtually no visible inclusions and fired to the same colour, similarly the Devonian Marl of the Welsh Border and the glacial till derived from it produced very similar fabrics from widely spaced localities.