Sex determination from unknown skeletal remains is of vital importance in forensic
anthropology, and much research has been conducted in this field. One of the most
commonly used methods is to create discriminant function formulae from metric data.
It is widely argued in the literature that these formulae are population-specific,
as various populations differ with regard to general body size and degree of sexual
dimorphism. However, the population of origin may not always be known, and formulae
do not exist for all possible population groups. The aim of this study was to assess
whether it is really necessary to develop population-specific formulae for pelvic
measurements. For this purpose, skeletons from three different populations namely
Greeks living on Crete (n=193), South African whites (n=200) and South African blacks
(n=199) were used. Seven standard measurements from the innominate bone were employed
and discriminant function formulae developed for each population separately, and then
for all three populations combined. Classification accuracies indicate that very little
is gained by keeping the populations separate. For example, in a stepwise calculation
using all measurements the overall classification accuracy was 94.5% for the combined
group, and 94.8%, 94.5% and 94.5% for the Greeks, SA whites and SA blacks respectively.
In a direct analysis using pubic and ischial length, the overall combined accuracy
was 89.8%, versus 89.3%, 90.0% and 89.0% for the three separate groups. When only
the acetabular diameter was used, the corresponding figures were 82.5% versus 84.1%,
81.6% and 83.5%. These results indicate that population-specific formulae may not
be necessary for pelvic data, but it should be kept in mind that the pelvis is highly
dimorphic. More research is needed, also including other bones of the skeleton.