baer 
1914] A Structural Study of Caterpillars. 111 
far as it goes, but is incomplete. The following comparisons 
may be made between Lyonet’s, Lubbock’s and Berlese’s 
lettering: 
LYONET. LUBBOCK, BERLESE. 
Ce 82 CXxXXIXa 
D, E 1,4,5 139-140 
A On 140b 
a 22 CXXXIX 
(CdS al G 61, 62 CXXXIV 
B, 7; 5, etc. 35, 76, 80, 81 CXLI 
a, b 16517, 21 CXXXII 
& 19, 20, 26 CxXXXIla 
c 18. CXXXIIb 
Pebes Ue 
ip Wetec, (On le ete. 147 
u, V 57, 58 XL 
The whole arrangement is complex and only in a general 
way comparable to that of the other segments. C+ is interest- 
ing as extending well beyond the segment line, and in Cossus 
the length of two whole segments—being the longest muscle in 
the caterpillar. It also crosses the middle line in Cossus, but 
not in Noctua, where it is shorter. If Lyonet is correct there 
are no longitudinal muscles within the nerves, (except perhaps 
the aberrant A and C+ which presumably represent the great 
longitudinal muscles of the following segments); and some of 
the muscles are innervated from the subcesophageal ganglion, 
evidently belonging to the cervical system. The spiracle is 
supplied by the “bride epiniere’’ or so called sympathetic 
fibre, derived from the same segment, but it runs largely in the 
following one and seem to supply also some of its muscles, 
besides anastomosing with its first nerve. This indicates 
strongly that the spiracle originally lay on the incisure, as the 
rudimentary second one actually does, and that it has moved 
forward. For the same reason we see that the other spiracles 
have moved back, being supplied by the nerve of the preceding 
segment; and all the spiracles are accounted for. 
The meso- and metathorax are perfect counterparts of each 
other, and strongly contrasted with the other segments in 
structure. The few points of difference between them noted 
by Lyonet, are further reduced by the correction of a couple of 
misinterpretations, and particularly by treating the three 
bellies of his a as separate muscles. In discussing these, and 
the remaining segments we may use Berlese’s division of the 
