Body mass is an ecologically and biomechanically important variable in the study of
hominin biology. Regression equations derived from recent human samples allow for
the reasonable prediction of body mass of later, more human-like, and generally larger
hominins from hip joint dimensions, but potential differences in hip biomechanics
across hominin taxa render their use questionable with some earlier taxa (i.e., Australopithecus
spp.). Morphometric prediction equations using stature and bi-iliac breadth avoid
this problem, but their applicability to early hominins, some of which differ in both
size and proportions from modern adult humans, has not been demonstrated. Here we
use mean stature, bi-iliac breadth, and body mass from a global sample of human juveniles
ranging in age from 6 to 12 years (n = 530 age- and sex-specific group annual means
from 33 countries/regions) to evaluate the accuracy of several published morphometric
prediction equations when applied to small humans. Though the body proportions of
modern human juveniles likely differ from those of small-bodied early hominins, human
juveniles (like fossil hominins) often differ in size and proportions from adult human
reference samples and, accordingly, serve as a useful model for assessing the robustness
of morphometric prediction equations. Morphometric equations based on adults systematically
underpredict body mass in the youngest age groups and moderately overpredict body
mass in the older groups, which fall in the body size range of adult Australopithecus
(∼26-46 kg). Differences in body proportions, notably the ratio of lower limb length
to stature, influence predictive accuracy. Ontogenetic changes in these body proportions
likely influence the shift in prediction error (from under- to overprediction). However,
because morphometric equations are reasonably accurate when applied to this juvenile
test sample, we argue these equations may be used to predict body mass in small-bodied
hominins, despite the potential for some error induced by differing body proportions
and/or extrapolation beyond the original reference sample range.