118 T. H. MORGAN 



not be expected within the numbers that appear in the experi- 

 ment. The double cross-over is eosin long whose expectation 

 is also rare. In table 11 there are two lots, Nos. 16 and 20, 

 in which no males appear in this class, but as the expectation 

 calls for very few flies here, it is unlikely that this can be the cor- 

 rect interpretation. If it were the correct interpretation we 

 should get evidence of this third lethal when the red females of 

 No. 16 were bred, as in table 12. But here there is no evidence 

 of a third lethal, nor is there any in the following generations. 

 It seems unlikely, therefore, that there can be any such lethal 

 (i.e., one beyond miniature) involved in this experiment. 



On the other hand, if the supposititious additional lethal should 

 be between eosin and miniature, as shown in figure G, the effect 

 would be to lower the class red miniature or eosin long below ex- 

 pectation for one lethal. In fact, the percentages of these classes 

 are too low, but one would not be warranted, I think, from these 

 data, in postulating a third lethal to account for this numerical 

 discrepancy. 



It seems probable, then, that the high ratios are either extreme 

 cases of the 2:1 ratio, or that there is present some disturbing 

 element not yet detected. The former view does not seem 

 probable for the numbers are large. The second interpretation 

 is more plausible, especially when the fact is recalled that the 

 start was made from an exceptionally high ratio (5:1), and the 

 line maintained by selecting the offspring of the pairs that showed 

 the highest ratios. 



There is another possibility that should at least be mentioned. 

 The males hatch later than the females, and in the crowded 

 bottles the males string out for some days after the females 

 have ceased to hatch. In the present case there was some crowd- 

 ing, but the bottles were run to a finish, or practically so, in order 

 to get the full male count. That the low male ratio is not due 

 to this condition is shown by the fact that the non-cross-over class 

 (eosin miniature males) compares favorably with the correspond- 

 ing classes of females. 



