V EXPERIMENTS ON SELECTION 241 



this particular variation. This method offers the 

 advantage of opening a wider field to investigation, 

 and on the other hand, animals or plants which, 

 through culture have been brought to vary in any 

 direction, are more apt to vary in the one particular 

 direction which may be desired. Even in cases where 

 one particular variation is desired in order to create 

 a special variety, de Vilmorin rightly advises as 

 follows : " In order to obtain from a yet unmodified 

 plant, varieties of a determined character, I would 

 strive at first to make it vary in any manner, and 

 would choose, as progenitors, not the accidental 

 variation which would most nearly approach that I 

 am seeking for, but merely the one which offers the 

 largest departure from the type. In the second 

 generation, I would again select as progenitors the 

 plants which are most different from the normal type, 

 and, at the same time, from the progenitors selected 

 in the previous generation." 1 In fact, as Vilmorin 

 says, we must try to craze the plant, to make it vary 

 as much as possible, in all possible directions ; and it 

 is only when this result has been obtained that it 

 becomes advisable to select the variations in the 

 desired direction. 



We should always remember that plants and 



1 Note snr itn Projet d^ Experience ayant pour but de creer ntie 

 Variete d* Ajonc sans Epines. Loc. cit. p. 35. 



