Through the coalescence of a number of factors in the past few years much more information, of a sort usable by the present Project has become available. Nonetheless, past practice shades information available today. Older publications, for instance, almost invariably contained only selected lists of the samian recovered from fieldwork, mainly that related to site chronology (as this is what excavators and others wanted of their samian), and lacked quantitative data. Hence, there are major groups of well-stratified samian from earlier published excavations which cannot be readily used for comparison with newly excavated groups due to the absence of either full catalogues by context/phase or quantification. In their review of Roman pottery studies for English Heritage Fulford and Huddleston had highlighted the general lack of quantification of samian groups in the past as a serious problem requiring amelioration (1991, 38). It was recommended that all future samian reports should include comprehensive quantification (cf. Fulford and Huddleston 1991, 45), whether this is undertaken by a samian specialist or the person writing and co-ordinating the pottery report as a whole. It is only recently that the reporting of quantitative information on the presence of samian amongst assemblages has begun to become more or less routine (Willis 2004, section 5.3.7). A consequence is that for a number of types of site (especially forts) and specific areas the amount of quantitative data in the public domain is limited. Significantly, for instance, although there are seemingly many pottery publications for sites of the northern frontier, modern and quantified data are only available for a very small subset of recent excavations, a proportion of which remain to be published (Willis 2004, section 3.5.1; Evans and Willis 1997, section 8.1; Willis 1997b).
An on-going difficulty of the past two decades or so has been a lack of compatibility between pottery reports in terms of quantitative methods employed. This lack of compatibility hinders synthetic analysis. The present report employs quantitative data by weight, EVE and numbers of vessels. It uses sherd count data only rarely. Whilst this latter measure has advantages it is subject to the strong effect of taphonomic processes, particularly breakage rates, and is not preferred (cf. Orton 1989). A number of reports, however, include only information by sherd count. This information is not used here for any analytical studies. It is axiomatic that the use of like measures can enable valid comparative analysis; this is particularly advantageous with sites from which several samples have been recovered at various locations, for it facilitates the possibility of examining spatial variation across the settlementscape (cf. Willis 2004, 4.2.4).