Abstracts Alex Bayliss Institute of Archaeology, UCL Classification is fundamental to archaeological study. Numerical techniques are routinely applied to archaeological data of all periods and regions. This paper attempts to validate the results of several common methods of cluster analysis which have been applied to an incidence matrix of the presence and absence of 1114 foundry stamps on 3389 surviving medieval bells from England. The analyses attempt to identify groups of bells cast by particular founders or foundries and groups of stamps contained in their work-boxes. The results of the analyses are compared not only with each other, but with the grouping of bells and stamps known from written evidence which can be related to surviving bells and marks. This is of different types-inscriptions on the bells themselves, related documentary sources such as wills and law-suits, and heraldry. In all 1237 bells (36% of those in the incidence matrix) have some independent evidence of founder or foundry. Not only is the accuracy and effectiveness of different techniques in identifying these groups assessed, but consideration is given to the assumptions of the techniques and how these relate to the archaeological processes evident in this dataset. These results may be of interest to other archaeologists who use cluster analysis but work in areas where such independent evidence is rare or entirely lacking. | Back to Abstracts | Back to the Timetable | |