Transition analysis: a new method for estimating age from skeletons

Abstract

Introduction Estimating the ages of skeletons is an essential part of any osteologist's job, regardless of whether the skeletons come from forensic or archaeological contexts. The ages of juvenile skeletons can usually be estimated with minimal bias and an acceptable range of error, although even this endeavor is not entirely free from problems. But the situation is far worse when one is dealing with the skeletons of adults. Here a number of serious osteological and statistical problems plague the process of estimating an individual's age-at-death (see Bocquet-Appel and Masset 1982; Boldsen 1988; Masset 1989; Jackes 1992, 2000; Konigsberg and Frankenberg 1992, 1994; Milner et al . 2000). Among these difficulties is the tendency of age estimates to mimic the structure of the known-age reference samples used as standards of calibration (a problem often called “age mimicry”) and an inability to estimate the ages of older skeletons (those greater than about 50 years of age). The Rostock protocol outlined in the present volume represents a step toward a solution of these problems. In particular, the strategy of estimating the entire age-at-death distribution f ( a ) before trying to estimate the age of any individual skeleton is an important, if at first somewhat counterintuitive, innovation. However, the strategy has one serious practical shortcoming: the target population has to be large enough to provide good estimates of f ( a ). Many archaeological skeletal samples examined by paleodemographers are simply too few in number to proceed in this manner.