EFFECT OF SELECTION 141 



negative variation* from the proper type of the new 

 variety. In this way the novelty may not appear to be 

 very far removed from a high normal variation of the 

 original type. The behaviour of the progeny of the two 

 types, however types which might thus in themselves 

 be readily confused is entirely different, and a ready 

 means of distinguishing them is thereby provided. 

 Each set of offspring shows regression to its own proper 

 modal value ; so that the offspring of the novelty show 

 a further marked development of the new features, 

 whilst the offspring of an extreme normal variation 

 resemble the type of the original form more closely 

 than they do their own immediate progenitor. 



If new types are not produced among domesticated 

 productions by the action of artificial selection, and all 

 that selection can effect is to pick out definite novelties 

 when they occur, the analogy between natural selec- 

 tion and artificial selection breaks down, and a large 

 and important section of the evidence in favour of the 

 production of natural species by the action of natural 

 selection is destroyed. In the place of this explana- 

 tion de Vries would put the theory of mutation, ac- 

 cording to which new species arise by single steps as 

 definite novelties, just in the same way as we find that 

 domestic varieties are produced. More than this, de 

 Vries believes that he has discovered a set of new 

 species in the very act of originating from an old 

 one in this way, a discovery which affords the basis 



* /.., a variant belonging to a class situated some dis- 

 tance from the mode of normal variability of the novelty, 

 and on the side of it nearest to the mode of the original type. 



