A CENTURY OF LARVAL DESCRIPTIONS. 39 



classification based on this, which naturally includes certain absurdities, 

 as any one-character classification must do. Still the two types of 

 structure are of value, though subject somewhat to adaptation, since 

 the circle type tends to become the line type with the change in habits 

 from an enclosed-feeding to an exposed-feeding mode of life. 



The relative positions of the hair- bearing tubercles seems to be a 

 character of some importance, especially as regards the postspiracular 

 and subspiracular tubercles, or the prespiracular (of Sji/niviidac), 

 tubercles iv and v. These characters are but seldom subjected to 

 adaptive modifications. They show three types of structure which 

 divide the order someAvhat broadly. The lowest type has tubercles iv 

 and V in line or nearly so, or united. The character is somewhat 

 flexible in the Tineidae, as nearly all larval characters seem to be 

 in this primitive group, and occasionally one of the higher types is 

 indicated ; but it holds very generally. The butterflies and 

 Saturnians have a similar tubercular formula, but may be differentiated 

 by other marked characters. The second type is characteristic of the 

 Xoctiddac and allies, tubercle iv being behind the spiracle. In some 

 of the lower families {e.;/., Epiplcmidae and Laslncaniindae'''-) it is not, 

 or but weakly, shown, but it holds generally throughout this large 

 group. The third type is characteristic of the Sphinnidae, tubercle v 

 before the spiracle. It holds perfectly even the aberrant African 

 Sphingid, LophostethuR dwnnlinii, which is covered with spines like a 

 Saturnian, but is nevertheless a true Sphingid and not related to the 

 group that it superficially resembles. Indeed I know of no exception 

 to the structure in this comparatively small family. The tubercular 

 arrangement is, therefore, of some importance, and should be accurately 

 stated in descriptions. Unfortunately, it has been, till recently, very 

 generally neglected. Dr. Wilhelm Miiller first called attention to it 

 (1886) but he was interested to show that the same general arrange- 

 ment held throughout the order, rather than to find the differences in 

 detail which indicate groups. In describing the full succession of 

 larval stages, much stress should be laid on the first one, that after 

 exclusion from the egg. Scudder has remarked how markedly 

 generalised are the characters of newly-hatched butterfly larvte. This 

 is conspicuous in the butterflies, since the first stage has the structures 

 of a low moth type which are suddenly replaced at the first moult by 

 the full or nearly full characters of a highly specialised butterfly larva. 

 Nevertheless, the phylogenetic interval represented by the first ecdysis 

 is equally as great in many moths. It is somewhat strange that this 

 embryonic first stage may be lacking without the fact being of import- 

 ance in classification. Naturally it indicates a specialisation, and 

 usually a subfamily or family group ; yet this difference may occur 

 Avithin the limits of a single genus (''.//., Apatela), if the facts have 

 been correctly observed. Many interesting instances of the partial 

 disappearance of the primitive first stage by the crowding back of later 

 characters are shoAvn in the Ayrtiidcw. The two principal differences 

 between the primitive first stage and the later ones are : (1) The 



* I am glad to accept Mr. Tutt's correction to my suggestion that the position 

 of tubercles iv and v in the Lasiocampidae is secondary. Mr. Tutt claims it to be 

 primary ; now, if so, this is a point in favour of the value of the tubercles in classi- 

 fication, since it shows the Lai^iocanipidac to be a generalised group, a conclusion 

 reached on other grounds. 



