34 g J. ARTHUR HARRIS 



For total solids and deviation of sucrose from its probable value, 

 r gi == .471 .067 



Thus in both cases the correlation between total solids and sucrose is 

 moderately high, as is necessarily the case because of the fact that total 

 solids are to a considerable extent made up of sucrose. 



Quantity of sucrose is in both cases more variable than the total 

 amount of solids. The correlation between total solids and the rela 

 tive amount of sucrose is in both cases positive and of a substantial order 

 of magnitude. Beet juice with higher total solids contains both abso 

 lutely more and relatively more sugar than that of beets with low total 

 solid content. 



Another way of approaching the problem is to calculate the correla 

 tions between total solids and the coefficient of purity, which is often 

 determined. 



Nevada series, 



r, P = -340 .027 

 Washington series, 



r sp = .529 .062 



These values are also positive and substantial, but their interpretation 

 involves difficulties due to spurious correlation. 



Illustration p. Relationship between total number of spikelets and 

 number of sterile spikelets in wheat 



GRANTHAM and GROFF (1916) have given data for the correlation 

 between the total number of spikelets and the number of sterile spike- 

 lets at the base of the head in varieties of wheat. Let t = total spikelets 

 and s = = sterile spikelets. Then from their two tables I find, without 

 applying SHEPPARD S correction, 



Bearded varieties, 



rt == -635 .039, r, ? = -f .164 .063 

 Beardless varieties, 



r t == -542 .054, r t ^= -\- .232 .072 



For both groups of varieties there is a positive correlation between 

 total spikelets and sterile spikelets and between total spikelets and the 

 deviation of the number of sterile spikelets from their probable value. 

 Sterility is, therefore, not merely absolutely but relatively more fre 

 quent in varieties with larger numbers of spikelets. 



