122 



George A. Sacher 



Next we may ask whether Ufespan can be predicted more 

 accurately by a combination of brain and body weight than 

 by either of them alone. The answer to this question is given 

 by the multiple regression of lifespan on body weight and 

 brain weight (Appendix Id), 



X 



0'6S6z - 0-2252/ + 1-035 



(4) 



it! 1.0- 



.5 1.0 1.5 



MULTIPLE REGRESSION VARIABLE 



2.0 



Fig. 4. Relation of lifespan to multiple regression variable 



defined by Equation (4) or Equation (8). Symbols as 



defined in legend to Fig. 2. 



This regression removes 84 • 4 per cent of the lifespan variance. 

 This is not a significant increase over the 79 per cent removed 

 by regression on brain weight alone. However, the scatter 

 diagram (Fig. 4) suggests that the multiple regression has 

 further reduced the differences between groups. It is possible 

 that a more detailed statistical analysis, in which a separate 

 allometric relation is obtained for each of the major taxonomic 



