(1983).1 Even in this case, 18 percent of the stationary female population would be postreproductive. In both these life tables, which describe extremely high mortality, there are still substantial numbers of postreproductive women, and the life expectancy of women surviving to age 45 is still an additional 14 to 17 years. There are very few elderly women, however.
The Hassan-Weiss life table has a radically different shape than those in the Coale-Demeny model life-table system: child mortality is lower, and adult mortality is higher. In the Coale-Demeny system, the lowest life expectancy at age 15 for females is about 31 years, and that corresponds to about 41 percent of births surviving to age 15, for an overall life expectancy at birth of 20 years. However, the Coale-Demeny model life tables for life expectancies of 20 to 30 years are mainly extrapolations from life tables with life expectancies of 33 years or higher, and the method of extrapolation has been criticized as exaggerating the mortality of children and understating the mortality of adults (Bhat, 1987; Preston et al., 1993). Preston et al. (1993) have developed a new set of model life tables for high-mortality populations, which are better based empirically and which avoid the methodological problems of the Coale-Demeny tables at low life expectancies. These model life tables have a shape between that of the Hassan-Weiss and Coale-Demeny life tables, as shown in the third column of Table 11-1. At a life expectancy at birth of 20 years, 44 percent survive to age 15, with a remaining life expectancy of 27 years.
It is true that there are some analysts of some skeletal collections who conclude that mortality was higher than these three life tables suggest and that the proportion of female births surviving to age 45 was lower (e.g., Lovejoy et al., 1977; Weiss, 1973). Others, analyzing similar collections, conclude that life expectancy was substantially higher than these life tables suggest (see the many life tables described in Hassan, 1981). Hassan (1981:121) sums up his survey with the view that "It is premature, pending further evaluation of the age distribution of prehistoric skeletal populations, to affirm that prehistoric mortality was greater than that for ethnographic hunter-gatherers." Which leaves open the possibility that it may have been greater.
The leading demographic studies of contemporary hunter-gatherer populations are for the Aché and the !Kung. Howell (1979:116) concluded that the !Kung had a life expectancy at birth of about 30 years in the past, although in recent years it appears to have been much higher. She also found that the !Kung
It may be questioned whether these model life tables, based as they are on extrapolation from the experience of modern high-mortality agricultural populations, are appropriate for preagricultural populations. Hill and Hurtado (1996:192), for example, argue that they are not. However, they appear to overstate their case in this regard. They base their conclusion on the selection of a Coale-Demeny table to match infant mortality, rather than on the best overall fit. It is readily seen that within the west female family, matching on either e0 or e10, both of which indicate a level-8 model table with e0 = 37.5, gives a vastly superior fit to the level 13 (e0 = 50), which Hill and Hurtado (1996:217) use to draw their conclusions.