138 



Clyde E. Leightt 



TABLE 6. Summary of Differences in Correlation Coefficients for the Same 

 Characters in the Four Varieties Compared 



Characters correlated 



Difference 

 Difference P. E. difference 



Yield and height 030 ± .013 



\ield and number of kernels Ool ± .005 



Yield and number of spikelets 069 ± . 006 



Yield and weight of straw 126 ± .014 



Yield and weight of kernels 147 ± .035 



Weight of kernels and height 101 zh .035 



Weight of kernels and number of kernels 184 ± .044 



2.3 

 10.2 

 11.5 

 9.0 

 4.2 

 2.9 

 4.2 



The largest absolute difference between the correlation coefficients for 

 the different varieties occurs in the correlation between average weight 

 of kernels and average number of kernels per culm, where it is .184 ± .044. 

 The relative difference here, as mdicated by the result obtained by dividing 

 the difference by its probable error, is 4.2 times the probable error. The 

 second largest absolute difference is when jdeld is correlated with average 

 weight of kernels, the difference here bemg .147 zb .035, which is agam 

 4.2 times the probable error. The third largest absolute difference is 

 when jdeld is correlated with average weight of straw, the difference 

 here bemg .126 ± .014, which is just nine times the probable- error.' The 

 only other difference above .10 is that existing when average weight of 

 kernels is correlated with height, which is .101 ± .035, this being 2.9 

 times the probable error. The- difference in the correlations between 

 yield and number of kernels is not large m absolute amomit, bemg but 

 .051 ± .005, but, owmg to the small probable error, it amounts to 10.2 

 times the latter. The difference in the correlation between jdeld and 

 number of spile elets is .069 ± .006, which is 11.5 times its probable error. 

 This absolute difference is again small, but the relative difference is large 

 because of its small probable error. The difference between the cor- 

 relations of yield and height is both absolutelj- and relatively small, being 

 .030 ± .013. 



The greatest absolute differences, then, are seen to be found when weight 

 of straw and average weight of kernels are correlated with other characters. 



