AND ENVIRONMENT I7 



characters you would have found them correlated '. To 

 test such a statement, every wife was mated with the hus- 

 band on the next grave in the churchyard, and of every 

 husband we inquired was he insane or not and of every 

 wife was she phthisical or not. We found that there 

 was no relation between a husband and a wife from 

 different graves; there was also no relation between 

 insanity in the husband and phthisis in the wife — in 

 both cases the correlations were zero. Thus we see 

 that if we take things in which we should anticipate 

 no relation, none arises. 



Another character in which we should expect marked 

 marital relationship is that of alcoholism, and, as we have 

 seen, we undoubtedly find it ! 



Alcoholism, Husband and Wife. 



All Poor -70 



Prosperous Poor '^ 



Middle Classes -27 



Here again in the middle classes 0-27 probably represents 

 the mating of like with like, for it is about the value we 

 have found for other characters. But 070 represents the 

 * infection ' of drink as evidenced in the criminal classes. 

 Had the relation between husband and wife come out on 

 this scale for phthisis we should have had no doubt 

 about attributing two-thirds of it to infection. Having 

 shown that husband and wife do not indicate in a 

 marked manner the influence of infection in phthisis, 

 we now turn to the relation of parent to child. 



This problem has been dealt with by myself on Dr. 

 Rivers' data in a paper of 1907, and more recently by 

 Dr. Charles Goring, when discussing his observations 

 on criminals. If infection be the principal factor in 



