No. 517] HEAVY AND LIGHT SEED GRAIN 



selected from heads in the neighborhood of 11.5 cm. long. 

 We should have to select the shorter heads in order to 

 secure the larger grains. 



Fig. 3 shows the correlation existing between the aver- 

 age weight of kernel as subject, and the total height of 

 culm, including the head, as relative. The correlation is 

 still negative and amounts to 40 per cent. The mean 

 length of the culm is 61.74 cm. 



The regression coefficient of the length of the culm 

 relative to the average weight of grain is — 0.987. As in 

 the last example, the selection of grains weighing 30 mg. 

 and above would imply that they had come from culms 

 about .")(> em. long, or about 5 cm. below the mean. 



As a summary, plants with shorter culms, shorter 

 heads, and with a smaller number of grains, bear on the 

 whole grains of a greater weight. The opposite of course 

 is equally true. 



If the data had been taken of a pure strain of oats, of 

 the variety studied— of a number of plants that had come 

 within a few generations from a single mother plant- 

 then the correlations might have varied slightly from 

 those given. If data were accessible of another variety, 

 then we might suspect even greater deviation from the 

 figures given. Variation in the same variety from year 

 to year may lie expected. However, there is nothing to 

 lead us to believe, from an a priori standpoint, that the 

 <lata given would be essentially changed. 



If data were taken on oat plants grown in hills, then 

 we might get less decided negative correlations than those 

 given where the plants were grown under field conditions. 

 The variability of the number of grains, the height of 

 plant and the length of head might be less, while the vari- 

 ability of the average weight of grain might not be much 

 different. 



The following table shows the coefficients of variability 

 expressed in per cent, of the various factors that have 

 been discussed. 



The factor of the number of grains is evidently the 

 much more variable one and perhaps under hill condi- 



