Pottery and the Archaeologist by Martin Millett

Author:Martin Millett [Millett, Martin]
Language: eng
Format: epub
ISBN: 9780905853086
Barnesnoble:
Publisher: Taylor & Francis
Published: 1979-01-01T00:00:00+00:00

Fig. 14. Two dimensional configuration of Portchester Pits showing how the majority of the pits cluster in the centre, indicating that there is no chronological structure. Configuration from the output of an MDSCAL program. For explanation see p. 39

The pottery was thus divided according to the functional categories used above (p. 37) and the assemblages were then described in these terms, with an additional descriptor, the percentage of fine ware in the assemblage (based on weight) being used as well. This was added as Fulford was unable to suggest reasons for the marked variation in this percentage. These nine descriptors were used for each pit and the data for the seventy-three pits were used to calculate a similarity matrix using the Clustan 1C program on the IBM 360 Computer at University College, London (Wishart, 1975).

From this matrix a dendrogram was drawn (figure 15) using the hierarchic fusion (single-linkage) method (Doran and Hodson, 1975: 175–8). The program used then executed a series of iterative relocations (Doran and Hodson, 1975: 180–1) in order to overcome the tendency for chains of clusters to develop, and in this single-linkage method for individual assemblages to become attached to one cluster, when they are more similar overall to another cluster. Six runs of the relocation procedure showed that most of the clusters were comparatively stable on a high level of similarity (below 200 on fig. 15) but they combined in completely different ways at lower levels. It therefore seems realistic only to consider the interpretation of the clusters on the high level of similarity. These clusters of high similarity have been numbered A to Q and the symbol of reliability, based on sample size, is added. This shows that clusters K, M, N, P, and Q may be disregarded from the point of view of interpretation since they are all based on very small sample sizes. The relocation iterations show that the remainder of the clusters stay relatively stable, with some slight movement between clusters. This movement is limited to a small number of pits which tend to switch position between clusters.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.