170 



D. F. JONES. 



seeds from the top half and bottom half were counted separately. 

 The data from the entire spikes are brought together and given 

 in Table I. to present the amount of selective action shown. The 



TABLE I. 



THE AMOUNT OF SELECTIVE ACTION SHOWN BY MAIZE IN FIVE POLLEN 



MIXTURES. 



figures from all the plants used with each mixture are combined 

 and are arranged in four classes in the form of a proportion. If 

 there were no differences in fertilizing ability the proportion 

 should be a perfect one within the limits of random sampling. 

 The white, smooth-seeded plants are designated A in the tables, 

 and the plants with yellow wrinkled seeds when uncrossed are 

 listed under the heading of B. The number of seeds resulting 

 from A pollen and from B pollen on A plants should be in the 

 same ratio as the numbers from the same two kinds of pollen on 

 B plants irrespective of the amount or viability of each kind of 

 pollen in the mixture. As in the numerous cases previously re- 

 ported such is clearly not the result obtained, as there are much 

 fewer cross-fertilized seeds than would be expected if fertiliza- 

 tion took place at random. The selective action is in favor of 

 the homogeneous union as was found before and the differences 

 are marked. The last column gives the deviation from the closest 

 calculated perfect proportion based upon the per cent, of cross- 

 fertilized and self-fertilized seeds on each type of plant. The 

 maximum deviation possible is 50 which would indicate complete 

 functioning of each kind of pollen on its own flowers, but not at 

 all on the others, or vice versa as the case might be. The devia- 

 tions actually range from 12 to 41 in favor of self-fertilization, 

 in each case, and show a very high degree of selective action, 

 amounting in some cases to almost total inability of the extrane- 



