104 THE AMERICAN NATURALIST [Vol. L 



supposed that selection was a cause of favorable varia- 

 tion. Instead lie repeatedly pointed out that the funda- 

 mental problem behind natural selection was that of the 

 cause of the variations which selection preserved. That 

 problem remains to-day practically in the same condition 

 that it was left by Darwin. We are no nearer, essen- 

 tially, now than we were then to knowing the cause of new 

 variations. The assertion that new variations are caused 

 by selection is the rankest kind of mysticism plus bad 

 logic. 



But if selection of the parents can not be supposed 

 the cause of new variations in the individual, then clearly 

 what selection does, and all it can do, is to change the 

 germinal constitution of a race or population by preserv- 

 ing those individuals in which new variations have ap- 

 peared, and multiplying them. This is exactly what has 

 been done in the hooded rat experiment, it seems to me on 

 Castle's interpretation of the case. In that experiment 

 every favorable variation in the many thousands of rats 

 has been preserved and the individuals bearing it have 

 been multiplied. Others have been thrown away. The 

 range of the character in the direction of selection has 

 been extended far beyond the original range. But would 

 it have been so extended, or could it have been, if the favor- 

 able variations had not appeared for selection, or if, hav- 

 ing appeared, they had not been heritable? Suppose one 

 started such an experiment with a character which was in 

 a stable condition and not varying. Take, for example, 

 the single comb of fowls, and attempt by selection from 

 a pure single-combed race to produce a stable rose-combed 

 race by selection alone. Prophecy is dangerous business, 

 but I do fancy one would be a very long time on that 

 job! Characters, so far as I can see, will be altered fol- 

 lowing selection just in proportion as they are varying 

 genotypically. The cause of the alteration is to be sought 

 in the cause of the variations, not in the selection only. 



I have for some time felt that probably the differences 

 in opinion between the selectionists, as represented by 



