Introduction

A majority of prior paleodemographic studies have focused on the estimation of population structure utilizing individual age range estimates derived from a variety of age indicators and compiled into a life table (Johnston and Snow 1961; Weiss 1973; Mensforth 1990). General population parameters from the life table are then compared among populations or with model mortality schedules (e.g., Coale and Demeny 1966). Recent research has demonstrated that age-at-death distributions derived from these types of age estimation method are biased as a result of an a priori assumption equating the age-at-death distributions of the reference and skeletal samples (Bocquet-Appel and Masset 1982, 1996; Konigsberg and Frankenberg 1992, 1994).

In this chapter, we will provide a case study based on an extension of the statistical methods detailed in Konigsberg and Herrmann (Chapter 11, this volume) using pelvic age indicator data from the large Archaic skeletal sample from Indian Knoll (15Oh2), Kentucky. This well-preserved skeletal series offers a unique opportunity to test these new methods. We compare the age-at-death distribution derived from this new approach with mortality data collected by several researchers from the Indian Knoll series. Our comparison illustrates differences between the earlier techniques, specifically life table based analyses, and our new method, which utilizes modeled hazard parameters and unbiased age estimates.

Indian Knoll history

The Indian Knoll skeletal series represents over 1100 individuals. The burial sample is one of the largest North American hunter–gatherer skeletal collections from a single site.