ME. Q-. J. EOMAISTES OK PHTSIOLO&ICAL SELECTION. 355 



of the reproductive system with which alone we are concerned, 

 and which must be extremely rare as compared with all the other 

 kinds o£ variation that it is the aim of breeders and horticulturists 

 to preserve ; for, third, it is never the aim of thesamen to preserve 

 this particular kind of variation. In view of these three consi- 

 derations, it is clear that we cannot expect to derive much evidence 

 of physiological selection from our domesticated varieties, further 

 than the general proof which these alFord of the primary im- 

 portance of preventing intercrossing with parent forms, if a new 

 varietal form is ever to gain a footing. No one of these domes- 

 ticated varieties could have been what it now is, unless such in- 

 tercrossing had been systematically prevented by man ; and this 

 gives us good reason to infer that no natural species could have 

 been what it now is, unless every variety in which every species 

 originated had been prevented from intercrossing with its parent 

 form by nature. For we have seen that even if the initial 

 variation, which, as a matter of fact, was in each case preserved, 

 happened to have been useful — and this supposition is, as we 

 have also seen, the reverse of true — it would still be so emi- 

 nently liable to extinction by intercrossing, that it is at least 

 doubtful whether its preservation could have been secured by 

 natural selection alone. Hence, although we cannot obtain much 

 direct evidence in favour of physiological selection from plants 

 and animals under domestication, we do obtain from them such 

 indirect evidence as arises from proof of the importance of pre- 

 venting intercrossing with parent forms. 



Again, as to plants and animals under nature, the particular 

 variation with which alone we are concerned would probably not 

 be noticed until it had given rise to a new species. In this 

 respect, therefore, the theory of physiological selection is in the 

 same predicament as that of natural selection ; in neither case are 

 we able directly to observe the formation of one species out of 

 another by the agency supposed ; and therefore in both cases 

 our belief in the agency supposed must, to a large extent, depend 

 on the probability established by general considerations. Never- 

 theless, although our sources of direct evidence are thus seen to 

 be necessarily limited, I shall now hope to show that they are 

 sufiicient to prove the only fact which they are required to proye, 

 namely, that the particular kind of variation which is in question 

 does occur, both in nature and under domestication. 



Although, as above remarked, the theory of physiological 



28* 



