No. 644] 



THE DURATION OF LIFE 



279 



its probable error, and must therefore be regarded as 

 statistically signijacant. Absolutely, however, the differ- 

 ence is small. It is equivalent to only 3.7 per cent, in- 

 crease of the expectation of life of the controls. In va- 

 riability in respect of duration of life there is plainly 

 no significant difference between etherized and contiol 

 groups. 



It is not entirely clear that the small difference be- 

 tween the etherized and control groups in mean duration 

 of life can be regarded as due to the influence of the 

 ether. An examination of the last two lines of Table II 

 shows that an entirely similar difference in the means 

 appears between the two control groups, which differ 

 only in respect of the time when they were taken from 

 the mating bottles, and without either having been ether- 

 ized. The difference in these two means amounts to 

 2.07 ± .52 days, a statistically significant and absolutely 

 slightly larger difference than that between etherized 

 and control groups. Again there is no significant differ- 

 ence in variability in the two groups. 



Altogether we shall be justified in concluding that there 

 is no evidence from these experiments that the occasional 

 etherization of Drosopliila to the extent necessary in sex- 

 ing and making matings alters the expectation of life 

 by an amount large enough to introduce any sensible 

 source of error into experiments on the duration of life 

 in this form, except possibly where the most careful and 

 accurate actuarial determinations need to be made. Then 

 it will be well to have this possible source of error in mind 

 and to plan the experiments in such way as to check it. 



Examining the results for the different etherized series 

 it is seen that the highest mean duration of life appears 

 in the group etherized once at 3^ days of age, and next 

 to this stands the group etherized once at 1| days of 

 age. Both of these give relatively higli mean values. 

 There also appears a definite, tliono-li not ]'>arti<Mi1ar]y 

 marked tendency for the variability in duration of life 

 to be greater in the groups which were etherized several 



