Abstract
The application of numerical taxonomic methods to archaeological problems is still very much in its infancy. A number of trial applications of various methods, to a wide range of archaeological data are clearly necessary.
The experimental research […] was undertaken to examine the applicability of three different methods of numerical analysis to the taxonomic study of adzes from New Zealand, and hopefully to a wider area. Of the three, factor analysis seems to be the most promising, for two reasons. It gave the most meaningful results when compared against the existing intuitive divisions of New Zealand adzes by noted authorities on the subject, and was able to demonstrate the regional variation among adze types which exists in New Zealand. This ability to demonstrate what was already believed to be the case must not be under-rated. The importance to the discipline of prehistory as a whole of the development of objective methods of analysis has been discussed at length above.
However, there is a much more fundamental reason for preferring factor analysis to cluster analysis. The simple plotting methods have been shown to be too limited in the number of variables which they can handle to be of great use. The other two methods can deal with large numbers of objects, and with a large amount of information about each object. However, there is a basic assumption involved in the use of cluster analysis which must cause it to be viewed with great caution in archaeological analyses. […]
[Extract from Chapter Six, Conclusions and Implications of the Study]