CLASSIFICATION OF LEPIDOPTERA. 105 



sary to enter in to here), we consider to be generally true. Theneuration of 

 the Micropterygids (Eriocephalids), Eriocraniids and Hepialids is, 

 perhaps, more generalised than that of any other Lepidoptera. Broadly, 

 on these lines, the neuration allows us to separate. the more generalised 

 from the more specialised superfamilies. When, however, one comes to 

 detail, i.e., to the consideration of the characters arising from the 

 modification of the neuration, we find the characters to be so variousl} 

 interpreted and applied by different authors, that, standing alone, the 

 neurational characters appear to be of very little value, witness 

 Hampson's groupings, by which on the character that the " Fore- 

 wing has nervure 5, arising from the middle of the disco-cellulars or 

 nearer 6 than 4, the nervures not arising at even distances around the 

 cell " the Cymatoplioridae, Notodontidae, Geometridae, Uraniidae, 

 Bombycidae, Ceratocampidae, Saturniidae, Sphinyidae, Dioptidae, Epip- 

 lemidae, Epicopeidae, Eupterotidae, Brahmaeidae and Ehopalocera,\ all 

 find themselves in one group. (4) Movable incisions of pupa. Chap- 

 man's pupal characters of movable segments, divide off sharply, and 

 with definiteness, the generalised, from the specialised, superfamilies, 

 the INCOMPLETE representing the former, the OBTECTJE, the latter, but 

 it is only in the details, such as those of the dorsal head-piece, the 

 maxillary palpi, etc., that we get any clue to the real relationships of 

 the superfamilies to one another, although the amount of incomplete- 

 ness of the pupa (i.e., the actual number of movable segments) affords, 

 in a comparative sense, valuable aid. (5) Hooks on proleys. The 

 arrangement of the hooks on the larval prolegs is largely associated 

 with a concealed or exposed habit of life, yet, with scarcely an exception, 

 the character is sound in separating the generalised from the specialised 

 superfamilies, and it is remarkable that even when a species belonging 

 to one of the specialised superfamilies, reverts to a concealed mode of 

 life, the prolegs do not revert to the generalised, but maintain the 

 specialised, proleg structure. (6) Larval tubercles. The arrangement 

 of the tubercles is remarkable from the fact that, more than any other 

 larval structure, they have undergone modification for protective pur- 

 poses. In concealed-feeding larvae, the tubercles have usually remained 

 simple, the setae often being suppressed until they form mere points on 

 the' chitinous button of the tubercle. On the other hand, in exposed- 

 feeding larvae, they vary from entire absence (where their presence 

 would interfere with the protective coloration adopted by the larva), to 

 raised warts bearing many setae, or they may form a prolonged spiny 

 base bearing several setiferous branches, or develop fascicles of urti- 

 cating spines, or hairs may arise from the normal base. In spite of 

 this, however, two characters remain fairly constant : (1) Tubercles j 

 and ii tend to form (by union or by the atrophy of i or ii) a single sub- 

 dorsal wart, or, on the other hand, tend to become arranged as anterior 

 and posterior trapezoidals. (2) Tubercles iv and v both remain as 

 sub-spiracular tubercles, or, on the other hand, v remains as a sub- 



* Annals and Magazine of Nat. History, Oct., 1894, pp. 254 et seq. 



t This must be really a very specialised character, for, sifting the families here 

 mentioned, one obtains the Cymatophoridae and Geometridae, the most specialised 

 of the Geometro-Eriocraniid stirps ; the Bombycidae, Ceratocampidae, Eupterotidae, 

 Saturniidae and Sphiiiyidae, the most specialised of the Sphingo-Micropterygid 

 stirps ; and the Notodontidae and Rhopalocera, specialised families of the Noctuo- 

 Hepialid stirps. 



