1898] A THEORY OF RETROGRESSION 401 



that A B C D represent a line of individuals ; then if D reverts to 



B, that is if D varies from his parent C in such a way that in his 

 ontogeny he represents the life-history of the race only up to the 

 point reached by B, omitting the additional characteristic of C, 

 it is evident from the point of view of heredity, that the series 

 becomes A B D ; or rather it becomes A B, since in effect, D is B. 

 C then disappears completely and for ever from the series ; and it 

 follows that, if the characters of C ever reappear in E, or any sub- 

 sequent member of the series, they must do so as a result of fresh 

 evolution, not as a result of reversion. It is necessary to emphasize 

 this point for on it my whole argument depends. If D, on the 

 other hand, varies in such a manner from C that after representing 



C, that is after recapitulating the whole of the phylogeny he reverts 

 back to B, then C does not disappear from the series. C will still 

 be represented in the ontogeny, and, if his characteristics reappear 

 in any individual at the end of the ontogeny, that is in the adult, 

 it will be as a result, not of evolution, but of reversion. As I have 

 already indicated, it is on such cases as the latter that Reversed 

 Selection works. Thus, when during the phylogeny any character 

 becomes useless and selection ceases, retrogression eliminates it 

 with a speed which is proportionate to the speed of evolution. 

 But, if it becomes worse than useless, then an additional factor 

 steps in to hasten the elimination. Reversed Selection then takes 

 advantage of such apparently atavistic, but really evolutionary 

 variations as cause an individual, after he has represented his 

 parent, to revert back again to a remoter ancestor. Moreover 

 Reversed Selection not only preserves such individuals, but also 

 eliminates all such individuals as have the worse than useless 

 characters in a greater degree than their parent, and thus pre- 

 vents them from influencing posterity. 



It would be well to illustrate the foregoing with a concrete 

 case. Suppose we plant seeds of those garden plants which I have 

 instanced as having undergone very swift evolution. In a great 

 number of cases the young plants revert towards the ancestral wild 

 type. Now I have enquired everywhere, and I have never heard 

 that the seeds of such a reverted plant, or of any of its descendants 

 have ever reproduced the cultivated type. This means that the 

 cultivated type has disappeared absolutely from the series. It will 

 never again be represented in the ontogeny, and could reappear 

 only as a consequence of fresh evolution, resulting from selection 

 as stringent as that by which the cultivated type was originally 

 evolved ; if it did reappear without fresh evolution it would be 

 because the reversion to the wild type had resulted not from true 

 atavism, not from a lapsing of the last steps of the ontogeny, but 

 from the false atavism on which Reversed Selection works. But, 



