Journal of Agricultural Research 
Vol. XI, No. 4 
II 4 
ME}AN LENGTH OF EARS 
In the first year (1905) we find that the means in length for any two 
of the strains are significantly different. The smallest difference that 
exists is 0.27 i 0.056. We thus note that the smallest difference is 
about five times the probable error. The average difference in the 
means is about nine times the probable error of any difference between 
two means. 
Having thus found in the 1905 crop significant differences when the 
differences are judged by probable errors, we very naturally inquire 
how persistently these differences are maintained and to what extent 
they are accentuated under continued selection for chemical composi¬ 
tion. In these considerations we must not lose sight of the differences 
in soil conditions, such as we have on Experiment Station plots, and 
Fig. i.—G raphs showing the mean length of the ears in the four Illinois strains of Indian com for ii crops, 
1905-1915. 
differences in seasonal conditions—that is to say, it should be kept in 
mind in considering these results that the various strains have occupied 
different plots of ground in different years and that there has been no 
special attempt to maintain uniform or comparable soil conditions 
among these breeding plots. 
In figure 1 is shown graphically the mean values of length of ears in 
the four strains. It may be noted from these graphs that the order of 
magnitude of 1905 was maintained with slight exception till 1911. The 
year 1911 was an abnormal one characterized by lack of moisture at 
the time when the corn was much in need of moisture. The high- 
protein and the low-oil strains changed positions in 1912, and these 
two strains have maintained the new order through 1915. In 1915 the 
high-oil surpassed the high-protein and the low-oil strain, but it is not 
unlikely that this is only a seasonal fluctuation. While there have been 
