408 



Mr. F. Galton. 



[May 27, 



the percentage of the dark eyes is diminished from 26 or 27 to 23, 

 that is to about one-eighth of its original amount, the hazel element 

 in either case absorbing the difference. It follows that the chance of 

 a light- eyed parent having hazel offspring, is about twice as great as 

 that of a dark-eyed parent. Consequently, since hazel is twice as 

 likely to be met with in any given light- eyed family as in a given 

 dark-eyed one, we may look upon two-thirds of the hazel eyes as 

 being fundamentally light, and one-third of them as fundamentally 

 dark. I shall allot them rateably in that proportion between light 

 and dark, as nearly as may be without using fractions, and so get 

 rid of them. M. Alphonse de Candolle has also shown from his data, 

 that yeux gris (which I take to be the equivalent of my hazel) are 

 referable to a light ancestry rather than to a dark one, but his data 

 are numerically insufficient to warrant a precise estimate of the 

 relative frequency of their derivation from each of these two sources. 



Heredity of Light and Dark Eye-colour. — In the following discussion 

 I shall deal only with those family groups of children in which the eye- 

 colours are known of the two parents and of the four grandparents. 

 There are altogether 211 of such groups, containing an aggregate of 

 1023 children. They do not, however, belong to 211 different family 

 stocks, because each stock which is complete up to the great grand- 

 parents inclusive (and I have fourteen of these) is capable of yielding 

 three such groups. Thus, group 1 contains a, the " children;" b, the 

 parents ; c, the grandparents. Group 2 contains a, the father of the 

 " children," his brothers and his sisters ; b, the parents of the father ; 

 c, the grandparents of the father. Group 3 contains the correspond- 

 ing selections on the mother's side. Other family stocks furnish two 

 groups. Out of these and other data, Tables III and IV have been made. 

 In Table III I have classified the families together whose two parents 

 and four grandparents present the same combination of eye-colour, 

 no class, however, being accepted that contains less than twenty 

 children. These data will enable us to test the average correctness of 

 the law I desire to verify, because many persons and many families 

 appear in the same class, and individual peculiarities tend to dis- 

 appear. In Table IV I have separately classified on the same system 

 all the families, 78 in number, that consist of six or more children. 

 These data will enable us to test the trustworthiness of the law as 

 applied to individual families. It will be seen from my way of 

 discussing them, that smaller families than these could not be ad- 

 vantageously dealt with. 



