A THEORY OF RETROGRESSION 59 



cultivated variety was undergoing evolution in 

 a certain direction, it, protected by man, was 

 undergoing degeneration in other directions, i.e. in 

 characters which were useful in the wild state, but 

 became useless in the protected cultivated state. 



It seems then that degeneration is due to 

 atavism — to a reversion to the ancestral type. 

 When reversion to a recent ancestor takes place, 

 the retrogression is comparatively small in amount ; 

 when a remote ancestor is reverted to, the retro- 

 gression is comparatively great. It follows as a 

 necessary consequence that rapid evolution is suc- 

 ceeded, on cessation of selection, by rapid reversion ; 

 for, under such conditions, reversion to any ancestor 

 results in much greater retrogression than would be 

 the case had the previous evolution been slower. 

 Moreover, evolution proceeds step by step, each 

 generation playing its part in it. But each genera- 

 tion does not necessarily play a part in retrogres- 

 sion, A son, skipping intermediate generations, 

 may revert at a bound to a remote ancestor. 

 A race-horse never springs from the loins of a 

 screw, the descendant of screws. But the son of 

 a race-horse, the descendant of race-horses, is 

 often a screw. Reversion, uncontrolled by selec- 

 tion, therefore, is always much swifter than 

 antecedent evolution. 



Every character had its beginnings. At one 



