A new approach to sexual diagnosis in past populations. Practical adjustments from Van Vark's procedure

Determination of sex is a key aspect in the study of past populations. Bias in skeletal sexing is well known, however, and depends upon the completeness of the skeletal remains and the representativeness of reference samples. In order to correct these difficulties, the authors propose practical adjustments to Van Vark's well‐known procedure. Two fundamental aspects are considered to reduce the bias of sex allocation. First, we propose to increase the effectiveness of sexing during the first step (primary diagnosis only from the pelvic bone). Second, to improve the final step (secondary diagnosis from the extra‐pelvic skeleton), we consider posterior probabilities as essential. These two practical adjustments are successfully tested in a sample of known sex and lead to a classification with 94.5% accuracy. These results significantly improve sex determination in archaeological samples and therefore contribute to a better understanding of past populations. Copyright © 1999 John Wiley & Sons, Ltd.