84 THE PRINCIPLES OF HEREDITY 



served by artificial selection, as in the case of Ancon and 

 Manchamp sheep. 1 But the selection is then extraordinarily 

 severe. The breeder endeavours by closer interbreeding to 

 protect his chosen abnormality from the regressive action of 

 sexual reproduction, and only thus can succeed in preserving 

 it. Under conditions of Natural Selection, lirparental repro- 

 duction insures that all evolution shall be on lines of small 

 variations, not on lines of great abnormalities. 



145. We have now conclusive evidence that parthenogenetic 

 species vary ; but, though the contrary has been too hastily 

 inferred, it is extremely probable that they do not vary to 

 very much more than half the extent that is normal in bi- 

 parental reproduction. It is likely that their progressive 

 variations are as numerous as in the latter case, but all the 

 scanty evidence available points to a comparative lack of 

 regressive variations. Thus the receptacula seminis of various 

 parthenogenetic species (e. g. Cypris reptans) appear to have 

 undergone no regression, though in some cases innumerable 

 generations have elapsed since they were functional ; whereas 

 in other cases (e. g. Aphides) when sexual reproduction alter- 

 nates with parthenogenesis the receptacula have degenerated 

 in the parthenogenetic females. It is not to be supposed that 

 there is no reversion in the absence of bi-parental reproduc- 

 tion. The fact that development is by recapitulation insures 

 a certain amount of it. Nevertheless it is reduced to a 

 minimum. Indeed, considering the evident lack of reversion 

 among asexual forms, it is probable that bi-parental repro- 

 duction plays a part in the causation of reversion greater than 

 is apparent on the surface. When two individuals who differ 

 obviously in certain characters mate, and bear offspring in 

 whom the characters are diminished or lost to the extent 

 of the parental differences, the reason for the regression 

 seems clear to us. But we are apt to be puzzled when 

 characters are lost or diminished in offspring the parents of 

 which resembled each other. In such instances the resem- 

 blances may have been only apparent ; differences important 

 from a developmental standpoint may have existed, and yet 

 have passed unnoticed, or they may even have been qualita- 

 tive and inappreciable. As is well known, we are apt to 

 overlook considerable differences, especially in unfamiliar 

 forms, unless our powers of observation have been trained 



should be placed in the same category. (See Appendix B, also The 

 Present Evolution of Man, p. 49.) 

 1 Animals and Plants, vol. i., p. 104. 



