Snow ran the numbers through an algorithm that he had created based on a reference set of hands from people of European descent who lived near his university. Using several measurements—such as the length of the fingers, the length of the hand, the ratio of ring to index finger, and the ratio of index finger to little finger—the algorithm could predict whether a given handprint was male or female. Because there is a lot of overlap between men and women, however, the algorithm wasn't especially precise: It predicted the sex of Snow's modern sample with about 60 percent accuracy.
Luckily for Snow, that wasn't a problem for the analysis of the prehistoric handprints. As it turned out—much to his surprise—the hands in the caves were much more sexually dimorphic than modern hands, meaning that there was little overlap in the various hand measurements.
"They fall at the extreme ends, and even beyond the extreme ends," Snow said. "Twenty thousand years ago, men were men and women were women."