224 A Classification of Lepidojjterous Larvae. 



sticky and transpai-ent. A broad white band separates the trans- 

 parent venter from the green sides. Head of a whitish color, jaws 

 brown. Joint 2 transparently granulated and marked with brown. 



The Specialized Frenat^ or Macrolepidoptera. 



First Division (corresponding in a general way to the Specialized 

 Frenulum Conservers of Comstock). 



The most primitive form of arrangement of the tubercles is seen 

 in that great group including the Noctuida?, Notodontidag, and their 

 allies. The tubercles are simple, single haired. Tubercles iv and 

 V are well separated, iv tending to become situated just behind the 

 spiracle. The sette of vii are never consolidated into a distinct 

 wart, but the three hairs are scattered on the outside of the leg, or 

 on the chitinous leg plate, which is frequently developed. The two 

 subventral tubercles are thus v and vi, instead of vi and vii as in 

 the Microlepidoptera, and appear to be situated the upper anterior, 

 the lower posterior. 



Specialization begins in a multiplication of the number of hairs 

 on each tubercle; but right here there arises a new line of modifica- 

 tion. It consists in the presence of hairs arising from very minute 

 tubercles all over the body. We shall see this modification again in 

 several different lines of descent ; but it probably reaches its maxi- 

 mum among the Notodontidfe in Ajmtelodes torrefacta. Here it 

 has replaced the process of multiplication of hairs on the tubercles, 

 which seem to be still single haired. The result of this process in 

 our Apatelodes angelica and in the South American genus Rosema 

 is the production of a structure closely like that of the Lasiocampidae. 



Several different modifications of the legs among the most gener- 

 alized Macrolepidoptera may now be noticed. The first consists in 

 the abortion of the two anterior pair of abdominal legs as seen in 

 the lower Xoctuidae, followed by the complete removal of the three 

 anterior pair as in the Geometridte, resulting, of course, in a loop- 

 ing method of progression. It may be noticed, in passing, that this 

 modification is much more strongly presented in certain Noctuids 

 in the first larval stage (embryonic larva) than later in life, which 

 might lead to the conclusion that the Noctuidas were descended from 

 geometrid-like ancestors, if we regarded the first larval stage as 

 always representing a generalized condition. This conclusion would 

 be, evidently, the reverse of the truth, as the specialization in the 



