24 ON THE ORIGIN OF SPECIES 



of reversion to ancestral characters, if all the domestic 

 breeds have descended from the rock-pigeon. But if 

 we deny this, we must make one of the two following 

 highly improbable suppositions. Either, firstly, that 

 all the several imagined aboriginal stocks were 

 coloured and marked like the rock-pigeon, although no 

 other existing species is thus coloured and marked, so 

 that in each separate breed there might be a tendency 

 to revert to the very same colours and markings. Or, 

 secondly, that each breed, even the purest, has within a 

 dozen or, at most, within a score of generations, been 

 crossed by the rock-pigeon : I say within a dozen or 

 twenty generations, for we know of no fact counten- 

 ancing the belief that the child ever reverts to some 

 one ancestor, removed by a greater number of genera- 

 tions. In a breed which has been crossed only once 

 with some distinct breed, the tendency to reversion to 

 any character derived from such cross will naturally 

 become less and less, as in each succeeding genera- 

 tion there will be less of the foreign blood ; but 

 when there has been no cross with a distinct breed, and 

 there is a tendency in both parents to revert to a char- 

 acter, which has been lost during some former 

 generation, this tendency, for all that we can see to 

 the contrary, may be transmitted undiminished for an 

 indefinite number of generations. These two distinct 

 cases are often confounded in treatises on inheritance. 



Lastly, the hybrids or mongrels from between all the 

 domestic breeds of pigeons are perfectly fertile. I can 

 state this from my own observations, purposely made, 

 on the most distinct breeds. Now, it is difficult, 

 perhaps impossible, to bring forward one case of the 

 hybrid offspring of two animals clearly distinct being 

 themselves perfectly fertile. Some authors believe 

 that long-continued domestication eliminates this 

 strong tendency to sterility : from the history of the 

 dog I think there is some probability in this hypothesis, 

 if applied to species closely related together, though it 

 is unsupported by a single experiment. But to ex- 

 tend the hypothesis so far as to suppose that species, 



