No. 557] INHERITANCE OF MAMMJE IN SWINE 



207 



in the third 1 : 4. The relative frequencies of these three 

 groups also look promising: 124 (0:N) :281 (1:4): 

 108 (1:2). The simple 1:2:1 ratio is approximated, 

 although the excess of the second group is a little high. 

 Records on next year's pigs are needed to prove the 

 soundness of this method of division. 



In a similar way the suppressed mammas show a strong 

 degree of inheritance. Both boars possessed the sup- 

 pressed mamma pattern. Sows with the suppressed 

 mamma produced 17 offspring with it and 49 without. 

 Sows without it produced 94 with and 350 without. The 

 ratios are approximately 1 : 3 and 1 : 3.7. Separating the 

 sows without the character into two groups on the same 

 basis as in the preceding case the following appears: 

 65 (1:3) from sows with the suppressed mamma, 388 (1:3) 

 from sows without the suppressed mamma and 56 (0:N) 

 from sows without the suppressed mamma. The frequen- 

 cies look like two 3 : 1 ratios combined in which the two 

 groups of three and one group of one are similar, but it is 

 not the normal ratio of 7 : 1. This also may be only ap- 

 parent, however, and another year's test is needed to 

 throw more light on the subject. It is beyond doubt, 

 nevertheless, that these two forms of variation are dis- 

 tinctly inherited and the manner of inheritance only is 

 that which remains to be demonstrated. 



Evidences of Segregation of the Two Types of Varia- 

 tion. — The grand-dams were plotted against their pigs 

 of the 1912 litters and the correlations were figured in 

 comparison with the dams. The results were not enlight- 

 ening, as the following table shows. 



Correlation between No. triangles in grand-dams and 



1912 pigs — .0064 ± .0624 



Correlation between No. triangles in dams and 1912 pigs .2241 ± .0592 

 Correlation between No. suppressed nipples in grand-dams 



and 1912 pigs — .0966 ± .0618 



1912 pigs ..... .... PP PP —.0052 ±.0624 



In the negative correlations the probable error is either 



