610 Bulletin American Museum of Natural History. [Vol. XXXIV, 
distinctly more variable with respect to the duration of their egg-larval 
period and only slightly, if at all, more variable with respect to their pupal 
period than those which succumbed early. As to the females, those which 
lived longest had been less variable in their egg-larval period and more 
variable in their pupal period than their weaker sisters. The discordant 
results, with respect to variability, taking this set of experiments as a whole, 
gives an additional indication of some unknown complexity influencing the 
outcome. The males best able to withstand starvation were less variable 
with respect to the length of the first posterior cell and the breadth of the 
wing than either those which succumbed early or the. general population, 
while those which succumbed early were slightly but not significantly more 
variable than the general population. In the case of the females the differ- 
ences were all insignificant. 
Normal Starvation 
Egg-larval and Egg-larval and Length Post. Cell 
Pupal Periods Pupal Periods and Breadth Wing 
_, General 
& Population +0.7470+ 0.0189 — 0.0561 +0.0624 | +0.8807+0.0143 
= Short lived +0.6591 + 0.0344 — 0.4383 +0.0728 | +0.9424+0.0103 
_ Long lived +0.8223 + 0.0194 | +0.2238+0.0827 | +0.8209+0.0315 
= | General 
= Population +0.7126 + 0.0205 — 0.0267 +0.0601 | +0.8449+0.0175 
= Short lived +0.6604 + 0.0293 —0.2757+0.0856 | +0.8254+0.0301 
F Long lived + 0.7823 + 0.0268 +0.1182+0.0789 -| +0.8014+0.0288 
Table 4. Coefficients of Correlation. See Table 2. 
In the “normal” experiments both the males and the females which 
lived longest (see Table 4) showed a distinctly higher correlation between 
the embryonic periods than did the short-lived ones. In the starvation 
experiments the survivors had a probably significant positive correlation 
between the embryonic periods while those which perished had a certainly 
significant negative correlations Apparently the zero correlation shown by 
the general population was caused by a mixture of two sorts which natural 
selection partly, at least separated. In the case of the anatomical characters 
no difference is shown by the females but the males which succumbed were 
less highly correlated than those which perished. | 
These results must seem unsatisfactory to those who look for hard and 
