Automatic Classification of Inspectional Categories: Multivariate Theories of Archaeological Data

Abstract

The application of extremely complex multivariate models of classification to subjective inspectional methods of categorization is analyzed in detail, with the widely used Texas system of dart point typology as a case study. The history of the development of the Texas dart point typological system is sketched. An attempt by Gunn and Prewitt (1975) to objectify the classificatory system by multivariate methods is criticized. The techniques applied were too idiosyncratic to the particular data set used to be of predictive value. Discriminant function and multivariate classification analysis are discussed in detail, emphasizing simple geometrical examples by which the major principles may be grasped. Suggestions for improvement are offered for those who wish to follow Gunn and Prewitt in constructing automatic classification schemes.