The numbers under "White" may be disregarded in 

 this connection since the distribution of long and round 

 pollen grains in this group shows a close approach to the 

 normal 3 : 1 ratio. 



Some of the numbers in the "observed" series are 

 above and some below the corresponding numbers of the 

 " calculated" series. That the calculated series approx- 

 imates the observed series is obvious, but there is no way 

 of determining the degree of this approximation. No 

 method has been proposed for making definite compari- 

 sons between such series of numbers. Without some 

 ••standard of comparison it is difficult to see that anything 

 is gained by resorting to gametic formulae to represent 

 the degrees of association between characters. 



A customary and direct method of comparing the de- 

 grees of relationship that exists between any two charac- 

 ters is to compute the coefficient of correlation or Yule's 

 "coefficient of association." 



In the following discussion Yule 's ' ' coefficient of asso- 

 ciation" (1900) is used. By this method the complete 

 independence of two character pairs is represented by 0, 

 complete association by 1. Intermediate degrees of re- 

 lationship are expressed by the intermediate decimals. 

 If the four classes of individuals are represented by a, b, 

 € and d, the coefficient of association is 

 (aX d)-(bX c) 

 (axd) + (bxc) 

 Since this coefficient can be computed directly from the 

 observed numbers the predication of gametic formulae as 

 a means of expressing degrees of association of charac- 

 ters becomes unnecessary. 



