524 



THE AMERICAN NATURALIST [Vol. XLYI 



plants than are those which do not deviate so widely 

 above or below the type. 



While the evidences in support of these conclusions are 

 fairly strong, it must not be forgotten that we are dealing 

 with a problem of great delicacy upon materials grown 

 under conditions such that the accidental (purely non- 

 selective) death rate must be quite high, and with records 

 not especially collected for our present purpose. Posi- 

 tiveness of assertion must therefore be reserved until 

 more critical evidence collected ad hoc is available. The 

 reader will note, however, that the conclusions here 

 recorded are based on many thousands of observations. 

 Although experiments are already under way, the gather- 

 ing of the data necessary for a more thorough investiga- 

 tion of the problem will be a long task, and it seems only 

 4 right that the results obtained incidentally should be 

 placed on record for the benefit of those who may have 

 opportunities for like observations. 



To me personally, the results were surprising. First, 

 I had doubted whether a selective mortality could be 

 detected by the methods used. Second, I had supposed 

 that if a differential viability were found, it would be 

 limited to a weeding out of the lighter seeds. This latter 

 presupposition was based on the fact that a positive cor- 

 relation had already been demonstrated 3 between the 

 weight of the seed planted and the number of pods pro- 

 duced, and a priori it seemed reasonable to suppose that 

 viability and capacity for forming plants with large num- 

 bers of pods would bear the same relationship to the 

 weight of the seed. Both questions deserve much more 

 detailed and refined study. 



Two questions concerning these results will be fore- 

 most in the mind of the biologist : {a) What is the signifi- 

 cance of the selective elimination for evolution? (b) 

 What are the underlying causes of the differential mor- 

 tality? 



