INHERITANCE IN ABNORMALITIES 441 



ency to resemblance between the stocks derived from the two 

 members of a pair, in respect to fission-rate, death-rate, and size. 



The question with relation to abnormalities may be attacked 

 by the same methods employed by Jennings and Lashley. The 

 problem presents itself as follows. From a certain number m 

 of exconjugants (forming m/2 pairs) a certain number n of 

 abnormal races are produced. In some cases both races derived 

 from a pair are abnormal; in other cases only one; in others, 

 neither. Is the number of cases where both races are abnormal 

 greater than would be expected if the pairing had no relation 

 to the distribution of the abnormalities? If so, this is evidence 

 that the pairing tends to make the two races alike in this respect. 



Jennings and Lashley ('13) give a formula for determining 

 the most probable number of pairs that will be found to be alike 

 in respect to any such character if the pairing has nothing to 

 do with the distribution. This formula is: 



k = 



{n + 1) (n/2 + 1) 

 m + 3 



in which the nearest integer below k is the most probable num- 

 ber of pairs in which the two members will be found alike in 

 the character in question; n is the number of cases (lines) in 

 which this character occurs, while m is the total number of cases 

 (lines of descent from exconjugants in this case). (This formula 

 holds absolutely only when n is even, but by obtaining. the result 

 for the two even numbers above and below the actual number, 

 if the latter is odd, the difficulty may be avoided.) 



In examining this matter for our case with respect to nor- 

 mality and abnormality, 17 of the 131 pairs of the first experi- 

 ment must be omitted from consideration, since in these the 

 characteristics of one or both members is unknown. This leaves 

 114 pairs to be dealt with, giving 228 lines of descent. In the 

 second experiment 3 of the 100 pairs must be omitted from 

 consideration, leaving 97 pairs to be dealt with, giving 194 

 lines of descent. In the third experiment there are 14 pairs to 

 be considered, giving 28 lines of descent. This makes 450 lines 

 of descent all together. 



