EVOLUTION 429 



^ 3. — Assortative Mating 



We have noted that if there is to be differentiation, 

 then a more or less effective barrier to intercrossing 

 between the selected components must be erected at an 

 early stage. While self-fertilisation and endogamy, under 

 which latter term we may include clan, caste, and class 

 unions, are obviously effective in maintaining differentia- 

 tion, the influence of like mating with like is slower or 

 more subtle, and requires delicate quantitative investigation 

 to demonstrate its existence. It is very probable that the 

 difficulties in the way of investigating its amount in any 

 form of wild life are insuperable — we can hardly face the 

 labour, to say nothing of the cruelty of destroying a 

 thousand pairs of swallows or sparrows at the mating 

 season, even if the comprehensive system of measurements 

 required could be rapidly and efficiently carried out. If 

 we exclude birds, it becomes in most forms of wild life 

 impossible or nearly so to identify the mates in any great 

 numbers. Thus we are largely thrown back for our 

 quantitative determinations of assortative mating on man. 

 When, however, we consider the strong feeling in his case 

 against blood marriages, the ease with which centre and 

 centre communicates, and the immense variety of individual 

 tastes which education and training produce, we might 

 almost despair of finding any distinct racial tendency to like 

 mating with like. Yet what is the result ? So far I have 

 only measured two characters, stature and eye-colour, yet in 

 both of them there is a quite sensible tendency of like to 

 mate with like. In fact Jmsband and wife for one of these 

 characters are more alike tJian uncle and niece, and for the 

 other more alike than first cousins. Such a degree of 

 resemblance in two mates, which we ma}- reasonably 

 assume to be not peculiar to man, could not fail to be of 

 weight if all the stages between like and unlike were 

 destroyed by a differential selection. 



During the last six years I have obtained measure- 

 ments of more than a thousand families, nearly all from 

 the middle classes. The data collected are the stature. 



