65 
a note on larvae of i\ ary via mondifera that needs confirmation, in which 
I wrote that the arrangement was trapezoidal . • •. on meso- and meta¬ 
thorax as on the abdominal segments. 
In Anthrocera (Zygjenids) and Adscita (lno), the arrangement is 
trapezoidal, i in front of ii, as on the abdominal segments; in these 
families, however, the anterior and posterior pairs are not widely 
separated, owing to the compressed nature of the segments. Dr. 
Chapman kindly gave me some young larvae of Heteroyynis paraduxa, 
but primary hairs are difficult to determine on the thoracic segments, 
owing to the presence of secondary tubercles. With the Pterorphorina 
or “ Plumes,” we get a divergence in position within the group, but I 
have examined so few larvae that I cannot as yet say how wide or far 
the rift extends. In Platyptilia yunudactyla the position of the dorsal 
tubercles on meso-thorax is transverse, with their bases conjoined, but 
on the meta-thorax, although still set transversely, the bases do not 
touch, but are still close together. Mimaeseoptilus phaeodactylm has 
them (the setie) set one in front of the other trapezoidally, i inner and 
anterior, ii outer and posterior; they are, however, close together, 
closer than is the case on the abdominal segments, and it is doubful if 
their bases, which are somewhat large, actually touch. Here I would 
suggest tentatively, until a more adequate quantity of material has 
been examined, that the characters in question (i.e.. the position of the 
dorsal tubercles on the meso- and meta-thoracic segments), are of con¬ 
siderable importance ; not as indicating relationship (in any but the 
wide general sense, of a tendency, probably throughout the Lepidoptera, 
to a reduction in the size and importance of the thoracic segments 
during the larval stage, with the consequent crowding or consolidation 
of tubercles), but as a guide, pointing out the distance to which the 
initial tendency of specialisation has been carried in different stirpes, 
or, stated somewhat differently, whether the groups quitted the main 
stream prior or subsequent to the time when this character had reached 
its zenith ; the fact of both positions occurring within the limits of a 
homogeneous group, such as the Plumes or the Hawk-moths appear 
to be, being evidence of an early division from the main stream, and 
subsequent independent development. Now I am of opinion that the 
stirps giving rise to the Lachneids, Saturniids, Endromids, Aylia, 
Ceratocampids, Hornby.v, Sphingids, etc., as well as that which gave 
rise to the Pterophorules, belongs to this category; the trend of the 
larvae in respect of the reduction of the thoracic segments not having 
reached its full strength when they branched from the main stream. 
Some of the Saturniid larvae bear on their meso- and meta-thoracic 
segments, forked processes, analogous to, but differing from, the 8th 
abdominal central horn in the direction of their forks or branches. In 
Hyperchina io, both the dorsal and supra-spiracular tubercles of the 
meso- and meta-thoracic segments bear processes or horns, the forks 
of which are set longitudinally, while those of the 8th abdominal horn 
are set transversely, pointing, as I think, to the direction in which the 
coalescence of the bases of the original tubercles occurred, the caudal 
horn being derived from the joined and elevated bases of i + i (both 
anterior trapezoidals) on the 8th abdominal segment, while the two 
dorsal horns or pillars which occur on both the meso- and meta- 
thoracic segments are evolved from i and ii on either side, and the 
forked character of the supra-spiracular derives its forks in all proba¬ 
bility from the original duplicate character of iii. 
