32 MAINE AGRICULTURAL EXPERIMENT STATION. I915. 



tions there is in each case a very marked decrease in the mean 

 index. This index is of the same order of magnitude with each 

 of the two selections. Further it is not essentially changed by 

 omitting the one disturbing line (No. 262) from the 1911 

 selections. 



From table ii it is seen that there is very little regularity 

 in the indices of the three selections. Thus with three plus 

 selections the mean index shows a tendency to lie slightly above 

 I, although probably not significantly. At least it may be said 

 that there is no evidence that three successive plus selections 

 have modified these pure lines in the direction of the selection. 



With three minus selections the case is somewhat different. 

 With each of the three selections the mean index is considerably 

 less than i, indicating that possibly the minus selections have 

 produced a real effect upon the pure lines. On going through 

 the individual indices it appears that the result is not due to a 

 few very small indices but that there is a real tendency for the 

 indices in these classes to be less than i. This result it will be 

 remembered falls into line with the conclusions drawn from 

 Tables 7, 8 and 10. 



Selection eor Characters other than Yield. 



In the preceding pages the only data dealt with have been those 

 for the yield of grain per plant. In order to see whether the 

 other characters upon which we have data behave in a similar 

 manner the records of so^me of these have been at least partially 

 analyzed. The characters to which most attention has been paid 

 are the height of plant and the number of culms. These show 

 but very little that is essentially different from the data for 

 yield of grain. On this account we will discuss the data for 

 height very briefly and omit the remaining characters. 



Tables 12, 13 and 14 show the same data for height of plant 

 that Tables 3, 4 and 5 show for weight of grain. 



