122 



THE AMERICAN NATURALIST [Vol. L 



TABLE I 



logical measurements. Attention has been called to the applica- 

 tion of this method of testing theoretical Mendelian ratios by 

 Harris (8). 



By calculation from the data given in Table I the measure of 

 agreement, or "P," is .418. "P" is a value ranging from to 

 1, proportional to the closeness with which the observed facts 

 agree with the theoretical. In this case in four times out of ten, 

 random samplings of similar data would give results deviating 

 more widely from the theoretical. A possible explanation for 

 this rather wide disc repancy will be given later. 



The children from matings of normal by abnormal are given 

 in Table II. 



A total of 448 children from 90 families is used. 232 children 

 out of the 448 are found to be defective, whereas 238 are ex- 

 pected according to the modified Mendelian ratio. 



As before, the criterion for including any family in the tabu- 

 lation is the production of at least one affected child. Only in 

 this way can the matings of heterozygous dominants, Nn (nor- 

 mal), with homozygous recessives, nn (abnormal), be distin- 



