378 
Journal of Agricultural Research 
Vol. XIV. No. 9 
Y/FLD / A/ 0£‘C/C/?A/7S 
coefficients in terms of their probable errors, there was practically no 
correlation between weight of seed and height of tallest culm at maturity 
in 1915 and 1917 and a good correlation in 1914 and 1916. The coeffi¬ 
cients of correlation between weight of seed and average height at 
maturity are o.o93±o.o38 in 1914, -0.099i0.028 in 1915, o.u8io.o3o 
in 1917, and 0.192 i 
0.024 m 1916. The co¬ 
efficients of correlation 
of the two characters 
range from 2.45 times 
the probable error in 
1914 to 8 times the 
probable error in 1916. 
Between weight of 
seed planted and aver¬ 
age weight of kernels 
harvested the coeffi¬ 
cients are — 0.062 i 
0.038 in 1914, 0.055 ± 
0.025 1916, 0.086i 
0.028 in 1915, and 
0.141 ±0.030 in 1917. 
While the coefficients 
are low in each of the 
four years, a slight cor¬ 
relation is indicated in 
1914 and 1916, and in 
1915 the coefficient is 
three times and in 1917 
4.7 times their respec¬ 
tive probable errors. 
The most significant 
correlation between 
the two characters occurred in 1917 when the seeds planted were 
selected from a line established through the selection of individual plant 
135 of the 1914 crop. Since the results indicate that the Marquis wheat 
used in this experiment was not homozygous for weight of seed, it can 
not be considered a pure line for this character. 
The coefficients for weight of seed planted correlated with yield of 
kernels per plant are o.o88±o.028 in 1915, o.i43±o.o38 in 1914, o.445± 
0.020 in 1916, and o.478±o.o23 in 1917. In 1914 and 1915 the coeffi¬ 
cients of correlation are low, 3.1 times and 3.8 times their probable 
errors, respectively, with no significant difference between them. The 
coefficients in 1916 and 1917 are considerably greater than those in 1914 
and 1915. 
Fig. 6.—Graph showing regression for weight of seed and yield of ker¬ 
nels per wheat plant in 1915. 
