Introduction

Population scientists concerned with long-term trends in human mortality ought to be interested in skeletal samples from extinct communities. Such samples are, in principle, the only possible source of information for most pre-industrial populations lacking written records – by far the most common kind of human community that has ever existed. Samples of skeletons provide two broad classes of information of potential interest to demographers and other population specialists: frequency counts of bony lesions that may reveal something about pathological processes active in the population, and data on ages-at-death from which age patterns of mortality may be inferred. Of these, the latter class of information has generally been deemed to be the less problematic. It has been assumed that skeletal age-at-death can be estimated well enough, albeit with some inevitable degree of error, to support a few crude but revealing statistics such as mean age-at-death, life expectancies, and age-specific mortality rates. And so for decades it has been considered perfectly acceptable to use skeletal data to compute “life tables”, the traditional demographic tool for investigating age patterns of mortality. All that is needed, in this view, are a few simple modifications of standard life table techniques, modifications that were laid down 30 years ago by Acsádi and Nemeskéri (1970:60–65).