234 Castle. 



characterized by a different meau size. Within each "line" he was 

 unable by selection to change the mean size. He concluded that the 

 asexually produced descendants of a single Paramecium constitute a pure 

 line of constant mean size, aside from environmental influences; and 

 that races characterized liy a different mean size arise only in con- 

 sequence of conjugation. 



But these conclusions appear of doubtful validity in view of the 

 work of Caxkens and Gregory (1913) who have isolated from the 

 asexually produced offspring of a single Paramecium strains differing 

 from each other in mean size more widely than the supposed '"pure 

 lines" of Jennings. If this is possible, we must conclude that genetic 

 variations may arise in asexual as well as in sexual reproduction, and 

 that they afford material for effective selection as to size. Horticultural 

 experience with asexual methods of plant propagation supports this 

 theoretical conclusion. Valuable horticultural varieties have repeatedly 

 arisen in the course of the asexual propagation of plants. 



So far as animals are concerned, "pure lines" have not been shown 

 to exist; so far as plants are concerned, we have only the beans of 

 JOHANNSEN said to be "pure" (that is devoid of genetic variation) for 

 size of seed. The evidence that genetic variation is wanting in this 

 isolated case is similar to that on wliich Jentjings relied to distinguish 

 pure lines of Paramecium, namely his inability to modify the racial mean 

 by selection. Whenever Johannsen was able by selection to modify 

 the racial mean, he assumed that he was dealing with a mixed "popu- 

 lation"; whenever he was unable to modify the mean, he assumed that 

 he was dealing with a "pure line". But the failure of JoHiVNNSEN in 

 certain cases to detect a change in the racial mean in consequence of 

 selection does not prove the non-existence of genetic variation in the 

 race, any more than my own failure to discover a pure line of animals 

 proves that such can not exist. Among Johannsen's beans somatic 

 variations may have been so much more numerous than genetic ones, 

 that his methods failed to demonstrate the genetic variations. I have 

 for some time desired to see the selection experiments repeated on the 

 same material witii a slight change of method, substituting the choice 

 of plants foi' fiiat of single seeds within the same line (as suggested 

 by Belling, 1912) in order to reduce the proportion of obviously somatic 

 variations. At one time 1 even essayed to make the repetition myself, 

 since no one else had, so far as I could learn, undertaken to do so. 

 Professor Johannsen had the great kindness to send me seed of one 

 of liis ))ure lines, but to my great regret I was unable to secure its 



