Some comments on coordinate-free and scale-invariant methods in morphometrics.

The unusual strategy for comparing biological shapes is to use some kind of superimposition of the two forms under study and then look at the "residuals" as the shape change. In this paper, I take a careful look at this general strategy and point out some subtle but inherent and important pitfalls. Additionally an alternative approach based on Euclidean Distance Matrix representation is presented. It is applicable to two- as well as three-dimensional objects.