EVOLUTION 417 



In-and-in breeding is one of the most powerful instruments 

 in the hands of the breeder, and this factor of artificial 

 selection seems replaced in nature by self-fertilisation and 

 by endogamy. Exogamy may be of value in providing a 

 greater range of variation, but in a period of differentiation 

 endogamy is invaluable. A tends to breed with A'', simply 

 because they come to birth and ripen in juxtaposition. 

 The eggs of an insect may be laid on one spot and hatch 

 out at the same time, the brood or family thus localised 

 keep together, and may ultimately breed together. 



Thus a differential selection accompanied by an 

 endogamous habit can cause two types in practically 

 the same locality to breed true, until intercrossing on 

 mechanical or physiological grounds becomes distasteful 

 or impossible. 



(iv.) I refer last to Mr. Romanes' theory of physio- 

 logical selection. According to this variations occur in 

 the reproductive organs or habits, so that the individuals 

 with these variations are mutually sterile. Thus in a 

 chance variation of reproductive characters an origin is 

 sought for the differentiation into species. According to 

 this theory in its narrow form, differentiation by natural 

 selection of characters not directly associated with repro- 

 ductive function would never be possible unless the two 

 types were accompanied by such chance variations in the 

 reproductive organs. This seems putting the cart before 

 the horse ; we require to seek why infertility accompanies 

 differentiation of type, and we are told that infertility 

 may be the source of differentiation in type ; but why, 

 then, should it be correlated with advantageous rather 

 than disadvantageous differences? Why are not the 

 chance reproductive variations distributed at random 

 between the two types ? 



In a later form of his theory Mr. Romanes recognised 

 that all differentiation requires the prevention of free 

 intercrossing. This is certainly true if selection is to be 

 secular and not periodic. The Darwinian theory of 

 evolution really requires then three factors : — 

 {a) Natural selection. 



27 



