214 Clyde E. Leighty 



whether plants are used as units or the culms of the same plants are used 

 as units, but that these constants will be slightly greater for the latter 

 method. The standard deviations and coefficients of variability will also 

 be greater when culms are the units. The biometrical results obtained 

 by the several investigators, whether they worked with culms or with 

 plants, are, however, comparable, with slight reservation, as far as this 

 factor affects the results. 



The second division of this study is a biometrical comparison of 

 varieties of oats. In this study, considerable difference is shown in 

 average yield of culm per plant. This is due to the larger kernels pro- 

 duced by certain varieties, since the numbers of kernels and of spikelets 

 are about the same for the different varieties. There are varietal differ- 

 ences in the height of culm. The average number of kernels per spikelet 

 is greatest in the Sixty Day and smallest in the Early Champion variety. 

 The proportion of straw to gTain differs in the different varieties. Con- 

 siderable difference is found in the amount of variability of different 

 characters of the varieties. The greatest variability, in all characters 

 but one, is found in the Welcome variety, while each of the others is least 

 variable in one or more characters. The coefficients of correlation are 

 usually fairly close together for the different varieties, but some differences 

 occur that may be due to varietal causes. 



The third division of this study is a comparison of biometrical constants 

 determined for oat plants gro^m in hills and in drills. Regarding all 

 characters here studied, the means are greater for plants grown in hills 

 than for plants grown in drills. The least difference in the means occurs 

 in the cases of average height of plant and average weight of kernels 

 per plant. There is greater variability in average yield of culm per 

 plant and average weight of straw in plants grown in hills, but much less 

 variability for plants so gro'um in average height and average weight of 

 kernels. The variability in number of kernels and number of spikelets 

 is slightly greater for the plants grown in hihs. Rather large differences 

 occur in the same variety between the coefficients of correlation determined 

 for the plants gro^^m in the two ways. Whenever large differences in 

 the coefficients of correlation occur, those for the plants grown in hills 

 are always the smaller in amount. The differences due only to the grow- 

 ing condition may amount to more than any varietal differences observed 

 in this work. The constants, then, obtained by different investigators 



