414 ON THE ORIGIN OF SPECIES 



It is, no doubt, extremely difficult even to conjecture 

 by what gradations many structures have been per- 

 fected, more especially amongst broken and failing 

 groups of organic beings ; but we see so many strange 

 gradations in nature, that we ought to be extremely 

 cautious in saying that any organ or instinct, or any 

 whole being, could not have arrived at its present 

 state by many graduated steps. There are, it must 

 be admitted, cases of special difficulty on the theory of 

 natural selection ; and one of the most curious of these 

 is the existence of two or three defined castes of workers 

 or sterile females in the same community of ants ; but 

 I have attempted to show how this difficulty can be 

 mastered. 



With respect to the almost universal sterility of 

 species when first crossed, which forms so remarkable 

 a contrast with the almost universal fertility of varieties 

 when crossed, I must refer the reader to the recapitula- 

 tion of the facts given at the end of the eighth chapter, 

 which seem to me conclusively to show that this sterility 

 is no more a special endowment than is the incapacity 

 of two trees to be grafted together ; but that it is inci- 

 dental on constitutional differences in the reproductive 

 systems of the intercrossed species. We see the truth 

 of this conclusion in the vast difference in the result, 

 when the same two species are crossed reciprocally ; 

 that is, when one species is first used as the father and 

 then as the mother. 



The fertility of varieties when intercrossed and of 

 their mongTel offspring cannot be considered as uni- 

 versal ; nor is their very general fertility surprising 

 when we remember that it is not likely that either 

 their constitutions or their reproductive systems should 

 have been profoundly modified. Moreover, most of the 

 varieties which have been experimentised on have been 

 produced under domestication ; and as domestication 

 (I do not mean mere confinement) apparently tends to 

 eliminate sterility, we ought not to expect it also to 

 produce sterility. 



The sterility of hybrids is a very different case from 



