342 MR. E. B. POULTON ON THE COLOUR-RELATION BETWEEN EXPOSED 
From this Table the duration of the stages can be approximately estimated as 
follows :—- 
Stages I. and II. 
Stage III. 
The whole period. 
hrs. mins. 
hrs. mins. 
Iirs. mins. 
1 
About 8 22 
About 15 50 
About 24 12 
2 
„ 5 15 
„ 20 20 
„ 25 35 
3 
„ 5 0 
„ 19 10 
„ 24 0 
6 
„ 5 15 
„ 20 20 
„ 25 35 
7 
„ 19 10 
. . 
8 
„ 15 50 
„ 17 0 
„ 32 45 
9 
„ 4 18 
„ 16 0 
„ 20 20 
10 
„ 16 0 
„ 19 45 
„ 35 45 
12 
„ 8 22 
„ 25 35 
„ 34 0 
13 
„ 20 20 
„ 14 30 
„ 35 0 
14 
„ 10 35 
„ 14 30 
„ 25 0 
The comparison between these pupae and those of Division B. was most carefully 
carried out, and the two divisions were compared together. The method adopted in com¬ 
paring the results of all careful experiments was as follows. The pupae were arranged 
side by side on a sheet of white paper, so that all their ventral surfaces -were illumi¬ 
nated by a strong east light which fell upon all the pupae at the same angle ; having 
thus decided upon their arrangement in the order of relative darkness or amount of 
gilding, they were all turned over, or the paper turned round so that the dorsal 
surfaces were illuminated in the same manner ; and the previous order was confirmed 
or modified, as the case might be; but on very nearly all occasions the two surfaces 
gave corresponding results, and the only exceptions were when the differences w r ere 
extremely slight. The results of the experiment tabulated above agree with those of 
Division A. in being destructive of the supposition that the desired terminal organ 
exists in the larval bristles. The figures seem to be very conclusive on this point. As 
to the length of the stages and of the preparatory period brought out by the above 
Table, there is little doubt that the larvae had in nearly all cases begun the period 
before being placed in the globes, and that the estimated length of the whole period 
and that of the two initial stages is far too short in nearly all cases. Further 
evidence of this suggestion will be adduced in the next division (D.). It is probable 
that the estimated length of the last stage (III.) is, on the whole, about normal, as the 
above 11 instances give an average length of 18 hours 16py min. for this stage. 
A few of the larvae seem to have passed through periods of about normal length. 
D. Another small number of larvae were also made use of to test the presence of 
terminal organs in the bristles. They were divided into two subdivisions as before:— 
(a) Of the nine larvae which were shorn on the evening of August 23, two had 
already quitted the food and were upon the roof or side. These were left in the lower 
of the two “ opal’ -covered cylinders with white-paper floors (described in III., A. and 
B.), the food was removed, and another larva was added which had been shorn on the 
