Relative variation in human proximal and distal limb segment lengths

The pattern of variation and covariation of proximal and distal limb segmen t lengths was examined within and between 20 geographically diverse skeleta l samples of modern humans. Analyses of variance-covariance matrices (VCMs) of logarithmically transformed (In) variates of humerus, radius, femur, an d tibia length were performed to test the following hypotheses: first, with in populations, the distal and proximal segments will have equal relative ( i.e., size-independent) variability. However, between populations, the tibi a is predicted to be more variable than the other segments. Tests of fit of computed VCMs to theoretical matrices by an iterative procedure (Anderson [1973] Ann. Stat. 1:135-141) reject the equal variance hypotheses, rather s uggesting that the relative variances of the distal limb segments are great er than are those of the proximal. Males and females differ somewhat in tha t within females, the distal segments of both limbs have equal variance, wh ile within males, the tibia has greater relative variance than the radius. The second hypothesis, regarding between-group variability, is somewhat sup ported in that between human populations, one cannot reject that the tibia has greater relative variance than the other limb segments. However, neithe r can one reject an alternative hypothesis that both distal limb segments ( tibia and radius) are more variable than the proximal segments. Differentia l growth allometry is explored, and likely plays a major role in difference s seen both within and between human populations. (C) 2001 Wiley-Liss, Inc.