550 



THE AMERICAN NATURALIST [Vol.1. 



last (F 2 ) plant. In the remaining 13 cases it was intro- 

 duced twice,— in consecutive generations in the case of 

 the group of 12 plants, with the skipping of a generation 

 in the remaining instance. Families were raised from 

 each of these 15 plants and doubles occurred in them all. 

 The numbers obtained, as was to be expected from the 

 more extended scale of the experiments, indicate a more 

 uniform proportion of singles and doubles than was ap- 

 parent in the earlier results. In many families the num- 

 bers clearly suggest a ratio of equality; in a few, how- 

 ever, there was a considerable excess on the side either of 

 the singles or of the doubles. No connection could be 

 traced between the proportion of doubles obtained and 

 the number of times doubleness was introduced into the 

 pedigree. The numbers in each family are shown below 

 where the 15 seed-parents are indicated by the capital 

 letters A to and the pollen parents by the small letters 

 a to h. 



How far these proportions are determined, as sug- 

 gested by Frost, 6 by a condition of partial selective ster- 

 ility can hardly be profitably discussed until further 

 microscopic investigations have been made, but that this 

 explanation forms a part, though possibly not the whole, 

 of the explanation appears highly probable. At the time 

 it seemed advisable to postpone further breeding until 

 wild material was available for comparison. Though 

 repeated efforts have been made to obtain seed of wild 

 plants, they have, in the case of P. violacea, been so, far 

 quite unsuccessful. In the course of 1912 and 1913 how- 



