Though the fossil record is still fragmentary and incomplete, the fossils are diverse and plentiful enough to present a very complicated picture of hominid evolution. Modern genetic techniques, which are very useful in reconstructing the phylogeny or kinship of the ancestors of living species, are useless here because these species have no living ancestors besides Homo sapiens and (with the exception of neanderthalensis and early sapiens) are too old to have left any DNA for analysis.

Like it or not, morphology matters, said Lynch, a biomorphologist who helped developed the mathematical analysis. For some problems you can't use genetics, and clearly the fossil hominid record is one of them.

The problem with morphology is that past methods of comparing bone forms have been largely subjective, and very hard to rigorously quantify.  "There has for a long time been a gap between the anatomy you can see on a skull and what you can actually measure, Lockwood observes.  These methods -- 3D geometric morphometrics -- help to fill that gap because they go to such a great level of detail.  And in this case, not only do they capture the anatomy, they show that the anatomy is a good guide to evolutionary relationships." 

One of the great things about this kind of technique is that you know if  you do the analysis right and you choose the right landmarks, you are exhaustively capturing all the data that exists for that bone, said Lynch. Intuitively, you can tell apart the temporal bone of a human and the temporal bone of a gorilla, but the beauty for us is that we found that our mathematical formulations were mapping those common-sense intuitive differences in a beautiful, quantitative way.

The fact that you can quantify the topography of the bones gives you the ability then to approach the differences both within and among groups statistically, said Kimbel.

This is the big leap, he said. It's finding a quantitative way to express the variation that we see and then having the ability both to analyze that variation with statistical robustness, as well as deriving phylogenetic information from it. Our next challenge will be to apply the method to the fossils themselves, and we're anticipating some fascinating results."