120 
Journal of Agricultural Research 
Vol. XI, No. 4 
some exceptions to this statement. At least we may state that the low- 
protein strain has for each of the 11 crops a larger standard deviation in 
length of ears than the high-oil strain. 
As shown graphically on figure 6, we find that on the whole the low- 
oil and the low-protein strains are more variable in circumference of 
ears than the high-protein and the high-oil strains. 
From figure 7 it is pretty clear that the low-protein and the low-oil 
strains are decidedly more variable in weight of ears than are the other 
two strains. When judged by probable errors, these differences are 
decidedly significant, and of such magnitude that we expect them to 
persist from year to year if soil and seasonal conditions are not extremely 
different. 
Fig. 7.— Graphs showing the standard deviation of the weight of the ears in the four Illinois strains of 
Indian com for n crops, 1905-1915. 
From figure 8 we note that the differences of standard deviations of 
number of rows of kernels for strains have not in general persisted well 
from year to year. For instance, in 1905 the standard deviation of the 
high-oil was greater than that of the low-oil strain by 0.31 ±0.06, but 
the next year it was less than that of the low-oil by 0.09 ±0.06. To 
have concluded from 0.31 ±0.06 that the high-oil strain has a larger 
standard deviation in number of rows of kernels than the low strain 
would have led us into error in 2 out of the 11 years. It may, however, 
be noted that the high-oil strain has each year of the 11 shown a greater 
standard deviation than the high-protein strain. 
Since the standard deviations, as shown in figures 5 to 8, do not show 
such an orderly arrangement as the corresponding mean values shown by 
