124 Experimental and Statistical Studies upon Lepidoptera 
A. Males. 
The pupae which produce perfect moths are found to be slightly shorter than 
those which fail to do so, and the former are slightly heavier also. In neither 
case, however, is the difference significant : there is no type selection in these 
characters. 
The data relating to the bust measures are of some interest. The perfect 
survivors are longer, narrower, and deeper than the imperfect survivors, and they 
are at the same time more slender in frontal and sagittal proportions ; in only the 
last two does selection appear with any degree of probability. It is highly 
important to observe that in all of these characters the perfect survivors bear the 
same relation to the imperfect survivors that the whole group of survivors does to 
the dead pupae. In brief, elimination of the second period appears to be a 
continuation of the earlier process of pupal elimination. Therefore the finally 
selected individuals are by far the longest, narrowest, and deepest, and at the same 
time the most slender ones of all that entered upon pupal existence. 
In the antenna, continued selection appears as regards length, but while 
broader organs occur in the perfect survivors the difference between these and the 
others is insignificant. Finally, the antennae of the perfect survivors are slightly 
more slender, though here, too, the difference is of no account. 
Although as we have seen there is no selection indicated in the type values of 
length and weight, when we compare the indices of variability in the two groups 
under consideration we find that the perfectly metamorphosing pupae are the less 
variable. Though this selection is but possible as regards the variability of length, 
a greater significance attaches to the difference between the variabilities in weight. 
In all the characters of the bust and of the antenna the perfect survivors are the 
less variable members of the whole group of survivors, but the difference between 
these and the other pupae is probably significant only in bust length, and certainly 
so only in antennal width. 
To summarize briefly, we find that type selection at the time of metamorphosis 
occurs with probability only in the proportions of the bust and in antennal length, 
while very little selection of variability is evidenced. 
B. Females. 
The pupae which produce perfect moths are slightly shorter, and slightly 
lighter than the others, selection being probable in the first case, and possible 
(almost probable) in the second. 
While in the males it was found that secondary elimination proceeded along 
the lines of pupal elimination, the pupae of the other sex show just the opposite 
relation. Specifically, the perfect survivors are shorter in the bust, and, though 
narrower as in the previous selection, they are less deep. In the proportionate 
measures especially they show the reversed nature of selection with the greatest 
clearness, for they are decidedly stouter in both frontal and sagittal planes with 
