MEASURE OF RESEMBLANCE OF FIRST COUSINS 



I I 



both series are : Health, Ability and Temper in adult cousins, and in sibships of school 

 children ; the following table gives the results for three classes : 



Table II a. 



Mean Cousinship "28. Half Mean Sibship -25. 



Treated alone these cases would show a definitely larger degree of l-eseinblance for 

 the cousinship, than for half the sibship, but this is not borne out for the whole 

 material. The differences also of the ages of the subjects, adults and school children, 

 and the methods of recording, by relatives and by school teachers, must also be borne 

 in mind. 



We consider on the basis of this first series that "25 is a good round working 

 number for the cousinship. This denotes on the assumptions of linear regression and the 

 equal variability of the cousins, that two cousins of an individual selected from unrelated 

 stocks (i.e. maternal and paternal cousins) will give the same probable value for the 

 character of an individual, as a brother of that individual with the same character as 

 the mean of the cousins*. On the other hand the accuracy of the estimate will not be 

 so great. In the first case it is o^ v 1 — ('5) 2 and in the second case o^ v 1 — ( - 25) 2 — ( - 25) 2 , 

 which measures the variability of the array; these are as 8'6G to 9*35. Thus the 

 prediction from the brother would be somewhat better than from two mutually 

 unrelated cousins. It is clear, however, that a knowledge of two such cousins may be 

 very useful indeed, especially if facts as to the sibship are not forthcoming. 



If we turn to other collateral relationships, the avuncular worked out for the eight 

 possible cases in eye colourt, is, as far as we know, the only one yet published. The 

 mean value of the eight cases is *265. We should accordingly conclude from this that 

 for purposes of inheritance a knowledge of the cousin is equally important with a 

 knowledge of the parental sibships. For example, there is no justification in medical 

 histories of lunacy for including the facts as to the parents' brothers and sisters and 



* Regression equation for 1 on 2 and 3, the latter being independent, is 



h, = 'Ji^i 1 H + 'J^ h 3 = r n h 2 + r u h 3 



<T., 0-3 



t Phil. Trans. Vol. 195, A, p. 114. 



if o-, 



•50 x J {l h + A,), if 



