148 A. FKANKLIN SHULL 



from the Baltimore pure line were mated, April 13, 1910, with 

 males from the New York pure line, and resting eggs obtained 

 April 14 and several following days. These eggs were put, with the 

 females, into an ice-chest at 8° to 10° C. on April 14, and left until 

 April 26, when they were again brought to room temperature. 

 On May 1, two females had hatched, one from each parent. 

 From these two females, the lines designated "Cross" in table 36 

 A and B respectively, were reared. The parts of the pai-ent (New 

 York and Baltimore) pure lines reared at the same time are given 

 for comparison. 



Here again the pure line derived from the cross yields in each 

 case more male-producers than do the parent lines, though the 

 differences are small, especially in the first two experiments. 

 The smallness of the difference is due, not to a lower proportion 

 of male-producers derived from the cross than in Experiment 

 XXXIII, but to a higher proportion in the parent lines than in the 

 former experiment. The percentage of male-producers in the 

 parent lines fluctuates considerably, and the delay in develop- 

 ment due to low temperature caused the cross to be compared with 

 the parent lines at a time of many male-producers in the latter. 

 Obviously, an experiment of such extent is needed that the error 

 due to these fluctuations may be reduced to a minimum. Such an 

 experiment is the following one, in which the reciprocal cross is 

 made. 



Experiment XXXV. On May 14, 1910, a female of the New 

 pure line was mated with a male of the Baltimore line, and 

 resting eggs obtained May 15 and following days. The eggs 

 were kept at room temperature until May 24, when one of them 

 yielded a female. From her, the line in table 37 designated ''Cross" 

 was bred. This line is compared with those parts of the New 

 York and Baltimore lines that occur simultaneously with it. 



That large fluctuations in the proportion of male-producers 

 in any hne do not vitiate this experiment seems probable from 

 the fact that the parts of the New York and Baltimore lines here 

 given include approximately the same percentages of male-pro- 

 ducers (10.8 and 15.9 respectively) as did those lines as a whole 

 (11.1 and 18.5 respectively), as shown in table 34. If these figures 



