Vaeiation and Correlation of Oats — Part I 67 



P When one compares the correlation coefficients for average height of 

 plant and average weight of kernels for the years 1909, 1910, and 1912^ — 

 which are .219 ±.029, —.023 ±.034, and .217 ±.032, respectively — 

 some information is obtained concerning the question of light and heavy 

 seed. For the two years 1909 and 1912 the coefficients are practically 

 the same, w^hile for 1910 the coefficient is negative and at the same time 

 it is so low that it shows no relation between these characters for this 

 year. These coefficients show that to some extent the taller plants tend 

 to produce kernels of larger average size for two years, while for the other 

 year this condition does not hold. If these facts represent the average 

 condition for oats for the years 1909 and 1912, were one to select large seed 

 from these crops for planting he would also be selecting plants that are 

 taller. These might have a tendency to produce heavier- j-i elding plants 

 the next year. On the other hand, if seed were selected from the 1910 

 crop, tall and short plants would be selected in equal numbers, and the 

 tendency on the whole would be for the plants of the succeeding crop to 

 represent the average conditions rather than to be larger because of the 

 large seed planted. 



The correlation coefficients for yield per plant and average weight of 

 kernels for the same three years are .149 ±.029, .035 ±.034, and .220± 

 .032, which are very similar to those for average height of plant and 

 average weight of kernels. For the year 1910 there is no relation as 

 measured by the coefficient of correlation between the two characters 

 in question, while there is some relation between these characters for 

 1909 and 1912, and, pro\'ided these facts are true for oats in general for 

 those years, by selecting larger seed one would be obtaining these from 

 the higher->4elding plants. On the other hand, when one selects seed from 

 the 1910 crop, plants having a tendency to \aeld high or low may be 

 obtained in equal amounts. 



Another correlation which is of importance in a practical way is that 

 between average height of plant and average number of spikelets per 

 culm per plant. This correlation for the two years is very high, which 

 shows that it is entirely possible to have tall plants which at the same time 

 will have a high average number of spikelets per culm. This is important, 

 since there is a good correlation between average number of spikelets 

 per culm and average number of kernels per spikelet. Takmg both these 



^ 1908 is omitted, since this represents average weight of kernels per culm rather than per plant. 



i 



