58 



H. H. Love and C. E. Leighty 



high. In all other cases, however, the correlation varies in the different 

 years from a negative correlation, sometimes of appreciable value, to a 

 comparatively high positive correlation. It is evident, therefore (1) 

 that average weight of kernels in oats has no necessary dependent relation 

 to yield, total and average number of kernels, average number of spike- 

 lets and of kernels per spikelet, and number of culms, and (2) that the 

 relation which does exist is largely responsive to environment. It seems 

 that when development is arrested by environmental conditions, yield 

 is reduced by reduction in number of kernels per plant, per culm, and 

 per spikelet, rather than in average weight of kernels and in number of 

 spikelets produced. 



Correlation of average number of spikelets 'per culm -per plant with other 



characters 

 The correlation between average number of spikelets per culm per plant 

 and average number of kernels per spikelet per plant (Figs. 63, 64), deter- 





? 





i-H 

 1 



(k 



rH 



1 



CO 

 .— 1 



3 

 1—1 



lO 



S 



7' 



I-H 



00 



s 





15-20 





1 





1 



3 



2 





1 



1 







9 



20-25 







2 



1 



6 



5 



6 





1 







21 



25-30 







4 



1 



8 



10 



16 



11 



7 



1 



1 



59 



30-35 





1 



1 



6 



12 



19 



25 



14 



14 



2 



3 



97 



35-40 







1 



4 



9 



15 



30 



33 



12 



6 



2 



112 



40-45 



1 







2 



10 



4 



10 



20 



16 



3 





66 



45-50 









2 





2 



5 



11 



6 



2 





28 



50-55 











1 



1 





4 



1 



1 





8 



1 



8 17 49 58 92 94 58 15 



400 



Fig. 64. 



■ Average number of spikelets per culm per plant, subject 

 Average number of kernels per spikelet per plant, relative. 

 Coefficient of correlation = .253 ± .032 



1910 



mined for two years, varies from .253±.032 for 1910 to.324±.027 for 

 1909. This is rather high and fairly constant. From this it may be 

 concluded that there is considerable relation between the number of 

 spikelets produced and the number of kernels produced in these spike- 

 lets, an increase in one being associated with an increase in the other. 



The correlation between average number of spikelets per culm per plant 

 and number of kernels per culm (Fig. 65) was determined for only one 



