120 Experimental and Statistical Stvdies vpon Lepidoptera 
Thus we see that in six of the eight characters under consideration, the 
surviving pupae evidence selection of type : that a typical condition, different 
from that of the dead pupae, characterizes those individuals which are physio- 
logically " fit " to survive the vicissitudes of pupal existence. 
When however we enquire if selection brings about the survival of the less 
variable individuals, the answer, though equally clear, is negative. Although in 
the depth of the bust, and in sagittal proportions, the dead pupae are more 
variable, they are less variable in the length, width, and frontal proportions of 
the bust. In none of these cases, nevertheless, does the difference exceed its 
probable error. There is in brief, no selection in variability. And when the facts 
relating to the antennae are examined, it appears that the survivors exhibit 
selection in length only with certainty, with possibility in width, and not at all in 
the variability of the proportionate measure. These statements will be found to 
be confirmed if the reader examines Table I. alongside Tables II. and III., thus 
comparing the unselected material with the selected. 
B. Females. 
After examining the facts relating to the female sex, we reach the remarkable 
conclusion that selection of a particular type occurs in all eight characters. In 
only one instance does the difference between the surviving pupae and the dead 
pupae fall below the error of the difference. Specifically, the survivors are longer, 
narrower, and deeper in the bust than the others, and they are also more slender 
in both frontal and sagittal proportions. Moreover their antennae are longer, 
wider, and stouter than those of the eliminated pupae. The actuality of selection 
is still further indicated by the fact that in all characters excepting the proportions 
of the antennae the variability of the former pupae is far less, and the significance 
is beyond the range of error due to random sampling. The same general con- 
clusions are reached if we compare the survivors with the unselected material. 
TABLE IV. 
Pupal Period. Selection in Type. Females. 
Cbaracter 
Mean ; Survivors 
Mean; Dead 
(S) 
Difference 
Error of 
Difference 
Proportion 
of 
5 : 6 
Selection 
Bust, Length 
„ Width 
„ Depth ... ... 
,, Frontal Proportions 
„ Sagittal „ 
19-50.39+ -0363 
12- 5949+ -0278 
1 3- 0007 ±-0i'87 
(;4-5727 + -0954 
G9-7727 + -0927 
19 1917 + -0443 
12- 8.388 + -0338 
13- 4250 + -0330 
67-0888 + -1079 
70-1111 ±-1082 
+ 0-3122 
- 0-2439 
+01757 
-2-5161 
-0-3.384 
0-0572 
0-0437 
0-0437 
0-1440 
0-1424 
> 5e 
> 5€ 
> 4f 
> 17€ 
> 2e 
Certain 
Certain 
Certain 
Certain 
Probable 
Left Antenna, Length . . . 
Width ... 
„ „ Proportions 
11 -6426 + -0231 
4-31 82 + -0102 
37-12.50 + -1002 
11 -38.33 + -0310 
4-1283 + -0126 
36-2667 + -0874 
+ 0-2.593 
+ 0-1899 
+0-8583 
0-0386 
0-0162 
0-1329 
> 6f 
> lie 
> 6e 
Certain 
Certain 
Certain 
