INTENSIVE SEGEEOATION. 313 



it maj be difficult to prove by measurement that there is any 

 difference in tbeir average character ; but on general principles 

 we may assume that, at least in some points, there is a slight 

 difference. It is evident that when the separated sections are 

 small there is more likely to be diversity in the average character 

 of the sections, and that, roughly stated, the probability of 

 divergence from this cause will be in direct proportion to the 

 variableness of the species, and in inverse proportion to the size 

 of the different sections. When a few stragglers form a small 

 colony in an isolated position there is the strongest reason to 

 expect that they will not be able to propagate the characters of 

 the species in exactly the same proportions in which they are 

 produced by the main body of the species, or by any other small 

 colony that is propagating independently ; and when the original 

 stock has been rendered highly variable by the crossing of some- 

 what divergent varieties, the degree of difference that will pro- 

 bably be presented by any two independent colonies will be 

 correspondingly increased. We must bear in mind that, while 

 specimens possessing an average character in any one respect are 

 always abundant, those perfectly representing the average in every 

 respect are rarely, if ever, found. Now, is it to be supposed that 

 any one, or any small number of these imperfect representatives 

 of a species will, if separated from the rest, transmit all the 

 characteristics of that species in the exact proportions presented 

 by the average character of the original stock? 



Mr. Francis Galton has conclusively shown* that in the children 

 of parents whose heights deviate from the average of the race to 

 which they belong there will be a similar deviation amounting on 

 the average to a certain fixed proportion of that presented by 

 what he calls the mid parentage. The mid-filial deviation in the 

 groups investigated by him was about two thirds of the mid- 

 parental deviation. There is therefore a regression in the average 

 character of the offspring toward the typical character of the 

 group. It must be observed, however, that this law can hold in 

 full force only where there is free crossing, otherwise no divergent 

 race could ever be formed by any amount of selection and inde- 

 pendent breeding. 



* See " Types and their luheritance," an address before the Section of 

 Anthropology of the British Association in 1885 ; also ' Natural Inheritance,' 

 p. 97. 



21* 



