200 A Classification of Lej)idopterous Larvse. 



smaller till it disappears, as may be seen in certain genera of the 

 Lymantriidog. This obliterates the fundamental distinction that I 

 have drawn between these groups. The distinction is still equally 

 good theoretically; but it fails in practice. For this reason I have 

 been in doubt about the position of the PyromorphidEe, Megalo- 

 pygidae (= Lagoid«), and Eucleidi« (:= Limacodidae), and I have 

 been obliged to give weight to the characters of the moths in locat- 

 ing these families. It is to be noted that the characters drawn from 

 the tubercles do not contradict the position which I have assigned 

 to these families, and I believe that I have correctly pointed out 

 the reason why the character of the tubercles fails. 



Several lines may now be taken by the increasing specialization 

 The tubercles may entirely disappear, either leaving the skin smooth, 

 or the whole body covered with pile or soft hair. The bases of the 

 tubercles may persist and be produced into fleshy processes, often 

 of great length, while the setae become aborted as we see in certain 

 Nymphalidge and Papilionidae ; or the setae may remain, and, becom- 

 ing glandular,^ secrete a poisonous, urticating fluid as in the Mega- 

 lopygidae and Eucleidae. Again, we may have the hairs greatly 

 complicated, developed into brushes, tufts, and plumes, or finely 

 feathered at the tips as in the higher Arctiidae, the Lymantriidae, 

 and Euchromiidse (;= Zygt^nidae of Kirby's catalogue). 



Perhaps the highest degree of modification exists in certain forms 

 in which the tubercles are partly retained, but tubercle i of each 

 side, approaching the dorsal line, has become consolidated with its 

 fellow of the opposite side, producing a row of dorsal tubercles ap- 

 parently unpaired. This seems to have taken place in two separate 

 lines of descent, and in a manner not quite parallel in the two. In 

 the first, namely that of the Saturnina (except the Lacosomidae) 

 and the Sphingida^, we have tubercles i consolidated dorsally on 

 abdominal segment 8, while ii are similarly consolidated on seg- 

 ment 9, tubercle i being absent on the other segments. In the 

 second line, which consists of a part of the Nymphalida^, tubercle i 

 is retained on all the segments, and we have a row of single dorsal 

 tubercles on abdominal segments 1 to 8 inclusive, without indication 

 of a consolidation of ii on the 9th segment.-' In both these groups, 

 a further modification of the tubercles is undergone, in that the 



' Perhaps the primitive seta is glandular. See Dr. A. S. Packard's many 

 papers on the glandular setae of young larvae. 



2 In some species, as Heliconia charitonia and Argynnis diana, this now is absent 



