ALLOMETRY AND THE ANALYSIS OF SIZE AND SHAPE IN HUMAN-EVOLUTION

This contribution has two main aims. The first is to assess the applicability of allometric techniques for purposes of predication and comparison, and the second is to test the possibility that the relationship between weight and stature was more significantly different in AL 288-1 (Australopithecus afarensis) and other Plio-Pleistocene hominids than it is in modern humans. It is argued that in the great majority of cases, the reduced major axis (RMA) is the preferred allometric technique. The reason for this is that the slope of the RMA is totally independent of the correlation coefficient. Both the least squares slope (LSR) and the major axis (MA) (which are both highly affected by the correlation coefficient) converge on it as the correlation coefficient approaches unity. It is also argued that the RMA provides the best estimate of the functional relation between the two variables in the analysis when the error variance is unknown, as is most frequently the case in allometric analyses. Furthermore, it is demonstrated that the MA produces a particularly spurious best-fit line when the correlation coefficient deviates from unity and the variance of the dependent variable (e.g., body weight) is particularly large in relation to the variance of the independent variable (e.g., stature). This has important implications not only for general allometric analysis but also particularly for the prediction of body weight from skeletal measurements. When RMA is used as the basis of inference and comparison, AL 288-1 (based on the best current estimates for stature and weight) has an inferred weight for its stature that would be highly unusual for modern humans, but similar to that observed for living African apes. OH 62 is similar to AL 288-1 in this respect. © 1992 Academic Press Limited.