The typological construction of biological races appeared to many to be an attractive methodology because it explained so much human variation without having to worry about confusing complexities, but it is those complexities that are the reality. This is the problem that confronted the practitioners of the ‘new physical anthropology’ as they tried to move away from typology. During the 1960s and 1970s two techniques were added to the armament of physical anthropology analysis. Both had their roots in the first half of the twentieth century.

OF SKULLS, BLOOD AND STATISTICAL ANALYSIS

Mention has already been made of Karl Pearson's biometric school and its mathematical approach to quantifying humanity. Pearson created something he called the ‘coefficient of racial likeness’, which was a mathematical formula that added up the differences in a series of measurements of two crania or two samples (Pearson 1926). The idea was interesting but ultimately unworkable because the measurements were intercorrelated (they measured similar things) and this made any conclusions unrealistic.

The solution to this problem was found by comparing the variation seen in the measurements, rather than looking at their averages. Hertha de Villiers (1968) chose to use Lionel Penrose's method (Penrose 1952), which separated out size variation and gave a better calculation of shape, while G. Phillip Rightmire (1970) and I (Morris 1984) worked with more complex multivariate techniques, which were only manageable with the advent of computers. None of these techniques identified race. What they did was to provide an accurate mathematical description (termed ‘biological distance’) of how crania differ in appearance. The most recent methods use geometric morphometrics. Images of bones are collected with a laser scanner and mathematical distance is computed from the three-dimensional spread of previously defined points – all done on a computer screen without having to touch a measurement tool to bone. This has proved invaluable for the detailed description of shape and is now commonly used in comparing differences between species where morphology has functional or adaptive importance.

But are these not just better methods of creating types? This is certainly a fair question as the techniques can indeed be used in exactly the same manner as the old typological measurements. For the most part, multivariate statistical analyses are being used in bioarchaeology research, which is interested in relationships between local populations, and race is not even part of the discussion.