62 



H. H. Love and C. E. Leighty 



i 



i 



6 



2 



3 



4 



5 



6 



7 



8 



9 



10 



11 



12 



13 



14 







1 



1 



1 













11 



4 



4 



2 



2 



2 







1 3 





12 



12 



16 



9 



6 



1 





1 





6 



10 



26 



23 



10 











6 



12 



16 



19 



5 



1 



1 







3 



9 



26 



15 



5 



2 



1 







4 



7 



21 



18 



13 



5 



2 







1 



1 



5 



2 

 1 

 1 



8 

 3 



9 

 1 

 2 



1 



3 

 2 





4 



25 



61 



78 



61 



61 



70 



23 



9 



5 



2 







1 



1 4 16 37 58 117 97 54 14 



400 



Fig. 71. — Number of culms per plant, subject 



Average number of kernels per culm per plant, relative. 

 Coefficient of con-elation = .4O8 ± .028 



1912 



of kernels per culm may on occasion be high, but is modified by environ- 

 mental conditions. 



Considered as a whole, the relation of number of culms per plant to 

 the other plant characters dealt with has been shown to be fluctuating, 

 generally to a considerable degree, being influenced by different environ- 

 mental conditions. The correlation is not especially high in any case, 

 except when total plant yield is considered. The conclusions may be 

 drawn, then, (1) that when conditions are favorable, the number of culms 

 will vary to a large extent directly with height of plant, average culm yield, 

 and average number of kernels, but that this relation may, on occasion, 

 entirely disappear; and (2) that the number of culms will vary directly 

 and constantly with the total plant yield; but (3) that the number of culms 

 and the average weight of kernels move practically independently of each 

 other under all conditions. 



Constants for 1910 compared with other years 

 The correlation coefficients for the year 1910 are worthy of special 

 consideration. It is seen that with one exception the only negative 

 coefficients found occur in this one year, five appearing in 1910. The 

 coefficients for 1910 are also noticeably lower, as a general rule, than 

 those of the other years. Of the twenty-two cases involved, there are 

 sixteen that are lower than those of the correlation coefficients determined 



