It is interesting to compare these figures to human populations. Austad (in this volume) takes the view that the evidence on human survival in preagricultural populations is dichotomous, with paleodemographic data indicating that ''humans until recently did not live into their postreproductive years except very rarely." while other data indicate much higher proportions of postreproductive females in the populations. He suggests that the disagreement in these estimates may be unresolvable.
A brief survey of the evidence may be useful. To begin with, we must operationalize "post reproductive." In developed countries, the mean age at menopause ranges from 47 to 50 years, but menopause comes some years after the effective cessation of the ability to bear children. In noncontracepting agricultural populations, the mean age at last birth is usually in the range of 39 to 41 years (Bongaarts, 1983:124-127). The mean age at last birth for the forest dwelling Aché hunters and gatherers is slightly higher (42.1 years; Hill and Hurtado, 1996:254) and for the !Kung is substantially lower (35.4 years; Howell, 1979:130). Nonetheless, many females in noncontracepting populations continue to bear children after age 40, so it is preferable to use age 45 for present purposes. By this age, about 70 percent of couples are sterile (Bongaarts, 1983:126), although often the sterility of the couple is due to sterility of the male rather than the female.
Evidence for survival in high-mortality populations comes from a variety of sources that include skeletal data, contemporary preagricultural populations who retain traditional life styles, high-mortality agricultural populations, and extrapolation from agricultural populations using statistical methods to generate model life-table systems.
There are many estimates of mortality based on skeletal (paleodemographic) data. These estimates have well-known weaknesses that arise from differences by age in the probability that a dead person will be represented by bones in the collection, difficulty in ascribing an age at death to the bones, distortions due to nonstationarity of the age distribution of the population giving rise to the specimens, etc. There is a very wide range in estimated life expectancies from skeletal data (Hassan, 1981), with some analysts reporting very low life expectancies, far lower than those estimated for contemporary preagricultural groups, whereas others estimate mortality at levels consistent with those for contemporary groups. Table 11-1 gives one estimate based on a combination of the model life-table methods developed by Weiss (1973) for preagricultural populations (here assuming that 60 percent of births survive to age 15), with an estimate of life expectancy at age 15 of 21.2 years presented as an average for a large body of paleodemographic data reviewed by Hassan (1981:118). The implied life expectancy at birth is 23 years.
The next column is based on a model life table with a life expectancy at birth of only 20 years and with an age pattern of mortality based on extrapolation from high-mortality agricultural populations analyzed by Coale and Demeny