CLASSIFICATrOflT OF GRACTIiARrA AND ALLIED GENERA. 83 



and inevitably fall victims in some degree to our more or less 

 narrow outlook. 



Another objection raised is, that we cannot classify by early 

 stages, because of our ignorance of them in so many instances. 

 This would be a valid objection were it the case that the demand 

 was to ignore the imago. Such an imputation is, however, pure 

 delusion. All that is proposed is that all the light that the 

 earlier stages throw on the relationships of species shall be used, 

 where we know them ; that what it tells us about a shall be 

 accepted, and not ignored, because it happens to be silent 

 about b. 



There is little doubt that the value of a knowledge of the 

 early stages for classification became very much neglected in 

 the eighties and early nineties, probably in consequence of the 

 enormous numbers of new species of exotic Lepidoptera con- 

 stantly coming to hand and being constantly described with 

 often no idea whatever of their earlier stages. The same causes 

 are no doubt still very active in the same direction, but some- 

 thing is being done in the opposite direction by not a few active 

 workers. 



The truth being that we require every scrap of knowledge 

 that we can get, about every species ; that the early stages are 

 quite as important as the imaginal, as illustrating relationships, 

 probably often more than less so ; but in any case, being further 

 information, they often guide us readily in cases where imaginal 

 indications are obscure, and must always be useful in checking 

 imaginal results and enabling us to see whether we have correctly 

 interpreted imaginal facts. 



It is because facts in the earlier stages are very pointed and 

 definite in separating Gracilariadse and Lyonetiadse from each 

 other and from other genera that I have selected them, partly to 

 illustrate this point, but chiefly to secure a better classification 

 of those families as a subject for this paper. 



The characters of the Gracilariadae that I propose to deal 

 with are two — one larval, the other pupal. 



The larval character is the very peculiar modification of the 

 mouth parts that exists in all the species in their first two 

 instars, continuing for further instars in some species. In all — 

 even in Phyllocnistis — changing suddenly at one moult to the 

 ordinary form. In Phyllocnistis, the third moult is of this 

 character, but the mouth parts are now useless for feeding, and 

 only available for spinning the cocoon. 



The pupal character consists in the movable or free segments 

 being reduced to the 5th and 6th abdominal, as in the obtect 

 pupa, but with the 7th also free in the male, and the habit of 

 protruding from the cocoon for emergence. It is, in fact, the 

 highest form of Incomplete Pupa, with the first four abdominal 

 segments fixed. I know of no other family, genus, or species 



H 2 



