90 H. D. GOODALE AND GRACE MACMULLEN 



begin to lay for the first time between February 14 and March 1 

 may be easily sorted out and classified as zero instead of mediocre 

 producers. We have not done this, however, in this paper as 

 they are too few to affect the ratios seriously. 



The results of assuming various division points are these. All 

 division points tried give ratios that fit the observed equally 

 well (compare table 5). The gametic classes in which the 

 individual breeders fall, however, shift with the change. For 

 example, in 1917-18, the various males belong to high-producing 

 classes with a division point at 30 eggs, but to low classes if the 

 division point is made at 60 eggs. 



DATA ON BARRED PLYMOUTH ROCKS, CORNISH, AND RECIP- 

 ROCAL CROSSES 



Although the application of the alternative theory to Pearl's 

 data, table 4, has ordinarily been made without difficulty, in some 

 instances, the fact that the data are grouped to fit a different 

 scheme has made it impossible to secure a good fit. An example 

 is given by male no. 558 which was mated to five females, all 

 referred by Pearl to Cornish females of class 1. They give the 

 ratio of 1 high to 9 mediocre producers. Although the 1 : 9 ratio 

 cannot be accounted for, it is clear that one female produced one 

 high and either one, two, three, or four mediocre daughters, the 

 other females producing mediocre daughters only. Such ratios 

 are accounted for by the alternative theory. The other cases of 

 poor fit are doubtless of the same character. 



In several instances it is possible to refer a male to more than 

 one gametic class, but this is not done unless the ratios are 

 identical. 



When Pearl modifies the ratios to accord with physiological 

 facts, the modified ratio is used, except when the modification 

 made does not affect the alternative theory, e.g., when some 

 zero producers are shown to be mediocre producers. In two 

 instances, viz., males no. 26 and no. 557, where the alternative 

 theory does not fit the observed ratios as well as Pearl's theory, 

 the deviations are not beyond the bounds of probability. More- 

 over, the deviation is in the direction that might be expected 



