118 



George A. Sacher 



40 



12 35 



o 



lU 



o 



< 



to 



UJ 



o 

 o 



o 



I- 



o 



< 

 cr 



< 



X 



o 



30 - 



25 - 



20 - 



15 - 



10 - 



5 - 



10 



10' 



I0> 



10^ 



10= 



10^ 



SIZE OF SAMPLE 



Fig. 16. Relation of characteristic oldest age to size of 

 initial cohort. The curves drawn are for a Gompertzian 

 life-table (see text) and for an exponential life-table. 

 The curve for the Gompertzian populations is repre- 

 sentative of the amount of variation of lifespan with 

 sample size that might be expected in populations of 

 domestic or zoo animals kept under good conditions. 

 The line for the exponential population is illustrative 

 of the relation of lifespan to sample size that would hold 

 for populations under very heavy environmental 

 pressure, such as small birds or rodents in the wild. 



Gompertz term* it follows that the lifespan is the preferred 

 statistic for the characterization of species longevity. 



* It can be argued that the life expectation should be preferred precisely 

 because it measures the response to environmental as well as intrinsic factors. 

 There can be valid reasons for preferring the life expectation in certain contexts, 

 and especially in the discussion of evolutionary or ecological adaptations. 

 However, the present paper is addressed to the question of intrinsic limitations 

 on length of life, and these questions are appropriately discussed in terms of 

 measures that are as nearly as possible invariants for the species, independent 

 of environmental influence. 



