20 Pearl and Bartlett. 



This gives a difference between observation and expectation of 

 0.375, certainly a reasonable result. The theoretical figure is extremely 

 close to the observed for the white starchy Fg kernels. 



Turning now to the sweets we have the following results. All 

 these kernels carry SS. Now, by hypothesis, in the presence of ss 

 we take: ^ ^ ,.,8 + ,.0. ^ ^ 



2 ■ -^ 



Ff = Ö = 9.02 

 Then we should expect to find in a random sample of either 

 yellow or white sweet kernels the following relations as to fat : 

 I FF = 6.65 



1 ff = 9.02 



2 Ff = 18.04 

 Sum = 33.71 



Whence = 8.43 = expected percentage of crude fat 



in Fo sweet kernels. 

 But 8.30 = mean observed percentage of crude 



fat in all Fo sweet kernels (yellow 

 obs. = 8.51, white obs. = 9.09). 

 This gives a difference between observation and expectation of 

 0.37, which is absolutely equal to that for the starchy kernels, though 

 relatively even smaller. 



Altogether the agreement between observation and hypothesis is, 

 for this character, so close as to indicate strongly that the hypothesis 

 is substantially correct, and that crude fat content behaves in 

 inheritance as an independent unit character. The further 

 test of the hypothesis will come as we apply it to other characters, 

 which we may now proceed to do. 



Let us consider next protein content. We will make exactly the 

 same hypothesis as before, namely that the presence of ss tends to 

 inhibit the expression of P (-- low protein content, the dominant 

 character). We then have : 



PP = 10.62 

 pp = 12.25 

 Pp =- 10.68 

 except in the presence of ss when 



„„ 10.62 + 12.25 



PP = ^ ^ = 11.435 



and Pp = 12.25 



