ABSTRACT

This chapter describes the application of a technique of multivariate statistical analysis to the classification of small mammal post-cranial skeletal material from the site of Bacon Hole, a cave in Gower. The traditional statistical method for the solution of such a problem is discriminant analysis. The linear discriminant analysis was carried out with the discriminant sub-programme of (SPSS) Statistical Package for the Social Sciences. The present SPSS discriminant procedure deals with missing data in the objects to be classified by replacing it with the mean of all the data. The small mammal species were identified from the teeth recovered at the Bacon Hole site. The British Museum had modern specimens of almost every species required for comparison. The fragmentary state of some of the bones from Bacon Hole meant that in many cases all the measurements could not be made. The general problem in all archaeological bone analysis, as archaeological material is most often fragmentary.