The results indicate that the secular changes in limb proportions cannot be explained by allometric responses to increasing size alone. The majority of variation among birth cohorts was found on the last PC, and that suggests that canalized development has been disrupted by the unique environment in which modern Americans now live. Static covariance matrices were homogeneous. Eigenvectors extracted from the secular covariance matrix showed important departures from static allometry, particularly a much stronger negative allometry of the humerus and a stronger positive allometry of the tibia. Shape analysis showed that relative humerus length decreased significantly over the time period examined and relative tibia length increased. The last principal component, which combined aspects of the brachial and crural index, showed the highest variation among birth year cohorts. Maximum lengths of humerus, radius, femur, and tibia were obtained from 19th and 20th centuries identified skeletons. Allometry was investigated on two levels, static and secular. Static allometry was defined as average allometry within 20‐year birth cohorts, and secular allometry as allometry among birth year cohorts. Allometry was assessed by extracting eigenvectors from covariance matrices of log transformed variables. Departures from allometry were examined using shape variables, and principal components of minor axes. (Publisher abstracts modified)