Relation of Lifespan to Brain and Body Weight 121 



The relation of lifespan to body weight is exhibited in Fig. 2, 

 with the points plotted on a log-log grid. The species are 

 divided into four groups, each comprising one or more orders, 

 and each group has a distinguishing symbol. No use is made 

 of this subdivision in the statistical analysis since the data 

 were not sufficient to justify a separate analysis by orders; 

 such analysis will be undertaken when more extensive data 

 are collected. 



It can be seen that there is a highly significant relation 

 between lifespan and body weight. The relation of log lifespan 

 in years (x) to log body weight in grams (y) is found by least 

 squares to be (see also Appendix lb) 



X = 0-198?/ + 0-471 (2) 



This regression accounts for 60 per cent of the variance of 

 lifespans. It can also be seen that the different groups are 

 stratified in the diagram, with the rodent lifespans lying below 

 the regression line, those for carnivores and ungulates more or 

 less evenly distributed around the line, and those for primates 

 lying almost entirely above. 



A similar diagram showing the relation of lifespan to brain 

 weight is displayed in Fig. 3. The least squares regression of 

 log lifespan (x) on log brain weight in grams (z) (Appendix Ic) 



X = 0-3252 + 0-684 (3) 



accounts for 79 per cent of the lifespan variance. This is 

 significantly greater than the variance reduction brought 

 about by regression on body weight, so it can be concluded 

 that brain weight by itself is a better predictor of lifespan 

 than is body weight. The superiority of brain weight over 

 body weight as a predictor is manifested by a reduced scatter 

 between the groups (shown by their clustering closer to the 

 regression line) and also by a decreased scatter within groups 

 (shown by the smaller dispersion of the individual deviations 

 from the mean deviation for the group). There is still evidence 

 of stratification, however, so that brain weight does not by 

 itself account for all the extractable lifespan variance. 



