MODIFICATION OF THE GERM-CELLS IN MAMMALS 223 



which Pearl ('17) has illustrated in his diagrams, figures 5 to 7, 

 pages 290 and 291. The chief difference being that we would 

 decrease the proportion of eliminated germ cells and increase 

 the proportion of defective and non-viable zygotes, and thus 

 emphasize the selection of individuals rather than of germ cells. 



A further consideration of Pearl's 1916 refsults as shown in 

 table 1, p. 676 ('16 b), may be used to argue in favor of our po- 

 sition. The 'prenatal mortality' column of this table when 

 compared with the 'dried in shell' column from 1915 records 

 (table 1, p. 244, '17) should show lower percentages according to 

 our interpretation of Pearl's expectation for an improved stock 

 from the alcoholic lines. Instead of this, in only one combina- 

 tion is the prenatal mortality lower. In both parents alcoholic 

 it has been lowered from 26.9 per cent to 11.11 per cent, and 

 here the postnatal mortality as we would expect is increased. 

 In the other cases dam only alcoholic, none of which were re- 

 ported for 1915 on account of the useless control male, gives 80 

 per cent prenatal mortality sire only alcoholic increased to 47.08 

 per cent from 36.6 per cent; sire and one grandparent, 46.84 per 

 cent; one or more grandparents, 46.02 per cent; all alcoholic 

 ancestry, 45.95 per cent, which is a considerable increase over 

 the 1915 records. The control of 1916 also shows a higher 

 prenatal mortality than that of 1915, though it is not stated 

 whether the same breed crosses are used in the two controls. 



The postnatal mortality of the 1916 control is, on the con- 

 trary, lower than the postnatal mortality of the twenty-two 

 'random sample matings' of 1915. 



While the total mortality for all the alcoholic groups is about 

 the same, 17.6 and 16.5 per cent, for the two seasons, the indi- 

 vidual combinations show wide variations. From both parents 

 alcoholic the 1915 postnatal mortality was 10.6 per cent, while for 

 1916 it rose to 25 per cent, sire only alcoholic fell from 21.1 per 

 cent, 1915 record, to 13.79 per cent, 1916 record. Sire and one 

 grandparent alcoholic gave a postnatal mortality of 28.38 per 

 cent, while the non-alcoholic postnatal mortality was 21.2 per 

 cent. 



Considering the numbers involved, the records from the prog- 



