THE AMERICAN NATURALIST [VOL.LI 



glumes ? This question can be answered by studying the 

 correlation between the flattening of head and hardness 

 of grain in a pure race. Since all of the No. 35 was soft, 

 this study could not be made for the soft wheat, but as 

 there were many plants of the pure macaroni which pro- 

 duced a greater or less proportion of soft grains, a com- 

 parison was possible. That the plants producing a large 

 proportion of hard grains on an average did not have 

 more flattened heads than those which produced a greater 

 proportion of soft grains is shown in Tables II and III 

 by comparing the "hard" with the ''intermediate" 

 groups of pure No. 1. This is perhaps better shown in 

 Table IV, where the distribution of the plants is made 

 with regard to their hardness per cent, and the ratio of 

 width to thickness of head. 



It should be noted from Tables II and III that there 

 are numbers of individual plants with low ratios of width 

 to thickness of head, but with high percentages of hard 

 grains. It is difficult to see how this could occur if the 

 low ratio was due simply to the lack of plumpness of the 

 hard grain. Moreover, races of hard ynacaroni wheats 

 occur which have approximately square heads, and there 

 are varieties of soft wheats {Little Club) with rather 

 strongly flattened heads. In 1916 there were a few cases 

 where hybrid races having low average ratios (W./T.) 

 also were rather high in average per cent, of hard grains. 

 All of the races in 1916 having ratios averaging as low or 

 lower than 1.35 and average percentages of 60 or more 

 per cent, of hard grains are given in the following table : 



It is thus seen that all of the races which on the average 

 violated the correlation in 1916 came from plants which 

 more or less markedly violated this same correlation in 

 1915. The cases given would indeed be hard to explain 

 on a basis of violation of physiological correlations but if 

 we are dealing with a genetic correlation, they may be 

 easily explained on the cross over" theory as used by 

 Morgan for reversal of linkages in the characters of Dro- 

 sophila. 



