EVOLUTION 497 



selection. Now we can answer this problem in two ways. 

 First, we may take any organ and determine whether the 

 death-rate is a function of the size of this organ measured 

 in any manner whatever. This would undoubtedly be the 

 best method were the results not apt to be screened by other 

 factors. We have, again, the difficulty of correlated organs 

 (see pp. 406-7 and 437) appearing. Even if we have hit 

 upon some organ upon which vitality directly and sensibly 

 depends, we have to meet difficulties due to growth and 

 to differences of age among the individuals dealt with. 

 Indeed, resistance to disease, health, and vitality in a given 

 environment may depend very largely on the complex inter- 

 relationships of a whole series of organs and characters. 



We accordingly turn to the second method in which 

 no attempt is made to select any particular organ, but we 

 take pairs of individuals having some general resemblance 

 in their whole complex of organs and characters, and 

 correlate their fitness for surviving under practically the 

 same conditions of life. Now pairs of relatives or mem- 

 bers of the same family are precisely such individuals. 

 If there were no selective death-rate, there would be no 

 correlation between the ages at death of, say, brothers. 

 On the other hand, if there were no non-selective death- 

 rate we might reasonably suppose duration of life would 

 be determined by the law of ancestral heredity, and we 

 should expect to find the correlation between the durations 

 of life of pairs of brothers about the .4 we have observed 

 in the case of stature, forearm, cephalic index, coat-colour, 

 eye - colour, etc. The reduction of the observed corre- 

 lation from .4 will enable us to determine the relative pro- 

 portions of the two death-rates. In order to determine 

 this point the ages at death of 1000 pairs of brothers 

 were extracted from the Peerage,^ as representing material 

 under practically the same environment. Minors were 

 excluded, as being often omitted from the record, so that 

 we are dealing practically with pairs of brothers who have 

 reached the age of reproduction. A correlation table 

 was then formed, precisely in the manner described on 



^ Royal Society s Proceedings, vol. Ixiv. p. 291, Beeton and Pearson. 



32 



