369 
On the metathorax of the green variety the subdorsal spots are 
present in double number, on the metathorax in single. 
Comparing the larval design to that of the imago, we may here 
remark in parentheses, that the last-mentioned pair of spots also 
occurs on the caterpillar of atropos and here maintains itself as the 
eye-spots of the cranium-image. 
The collection Katiensacn also contains a halfgrown convolvuli- 
caterpillar, I therefore found occasion to compare this with the 
fullgrown larva, and thus could convince myself that the light sub- 
dorsal maculae of the latter really are the remnants of a complete 
subdorsal line, originally stretching over the whole series of the 
body-segments, in the same way as the substigmal line. On the 
abdominal segments however it is periodically broken by the diagonal 
striae which, though rather inconspicuous, are yet present in complete 
order, and which, before each stigma, meet the segments of 
the substigmal line, thereby forming a triangular spot on each body-ring. 
But the chief difference between this halfgrown caterpillar and the 
fullgrown one is its uniform dark ground-colour, sharply contrasting 
with numerous small white oval knobs, which stand arranged in 
several rows on the eight annuli. On the above-mentioned light 
longitudinal lines these knobs occur in the same way. They do not 
make the impression of standing in any relation either to the dark 
or to the light specks, their own hue in fact is a much clearer 
white than that of the latter. So we can only suppose, that at the 
last moult they disappear, to be replaced by the (tolerably regular) 
white spots in the dark ground-colour. 
When we pass from the caterpillar of convolvuli to that of atropos 
and digustri, the design of these latter two is seen to correspond to 
that of the first in all such instances, as can be considered as secon- 
dary modifications of the original pattern; those parts of it on the 
contrary, which in convolvuli are found in the least modified condi- 
tion, having nearly vanished in the other two species. The same 
may be said of the thorax: here the process has led to total absence 
of pattern in atropos and ligustii. Upon the abdominal segments on 
the contrary the pattern is the same for all three, only differing in 
shades and in completeness. 
These facts undoubtedly offer new and valuable arguments for the 
supposition, that the absence of pattern is a consequence of obliteration ; 
the two latter species therefore having suffered stronger regression 
from the original condition than convolvult. 
Still in another instance the lastnamed species seems to show the 
more primitive conditions, viz. in the simple and few colours that 
