THE PHYLOGENY OF THE LEPIDOPTEROUS PUPA. 91 



the jaws. It is usual to regard these as an incomplete disappearance 

 of the larval jaws, but there is good ground for regarding them rather 

 as representing the pupal jaws of the Palaeo-Lepidoptera. We may 

 parallel them with the maxillary palpi of some pupae, where the larvae 

 have no such developed palpi on the one hand, and the imagines have 

 none whatever on the other, but they persist in the pupae by descent 

 from pupae that possessed them, not, as is often stated, from imagines 

 that had them, for, assuming the pupa to be descended in this way 

 from an imaginal ancestor, it would be able to follow the imago more 

 closely than it does and would lose the palpi pari-passu with the 

 imago. The tendency to pass characters from one stage to another is 

 too strong for the current explanation to be valid. Some hairy larvae 

 {e.g., Liparids) pass on hairs to the pupae, and, because they are advan- 

 tageous, they remain ; other hairy larvae no doubt tend to pass hairs 

 to the pupae, but not being advantageous they are eliminated as rapidly 

 as transmitted. The pupal maxillary palpus is descended, then, from 

 a pupa with a maxillary palpus, a pupa that possessed it for many 

 ages, no doubt because the imago had it, but which possessed it as 

 a transmissible character, and retained it as presenting no serious 

 inconvenience long after the imago, under severer selection, had lost it. 

 The pupal mandible is, therefore, more probably descended from a 

 pupal mandible than continued on from the larva, though no doubt 

 the tendency for the larva to transmit a mandible to the pupa, which 

 would by itself be ineffective, does assist in preventing natural selec- 

 tion eliminating it, a task difficult in itself on account of the ancient 

 lineage of the structure. 



The lower Neo-Lepidoptera, then, escape from the cocoon without 

 imaginal jaws and without pupal jaws, but do so by aid of the effete 

 pupal skin, by means of which a way, usually prepared in some manner, 

 is broken out of the cocoon by a beak, or wedge-shaped end to the 

 pupa, and the necessary force exerted by rough or spined surfaces on 

 the abdominal segments. This (physiological) character forms the best 

 general definition of a pupa of the Lepidoptera-Incompletae. There 

 are many subsidiary characters, structural and other, usually associated 

 with it, but at times wanting, whilst as rare and interesting excep- 

 tions, the definition would include such obviously obtect pupae as that 

 of Endromis versicolor. 



The most universal character of the pupa-incompleta is, that the 

 free segments of the abdomen, i.e., those that are free to move on their 

 fellows on either side of them, are always one more in the male than 

 in the female pupa. The 7th abdominal segment always forms part 

 of the terminal solid mass in the female, but is free in the male. The 

 total number of free segments varies from seven in a Nepticulid or 

 Cochlidid male pupa, to two in a female Gracilariid pupa. A very large 

 proportion of the pupae-incompletae have the 3rd, 4th, and 5th seg- 

 ments free, so that we may suppose this to have proved the most 

 advantageous number for the purpose of forcing the pupa out of the 

 cocoon. In those instances in which only 4 and 5 are free, practically 

 only the Gracilariids, some special circumstance must have dominated 

 the position. Since the large number of free segments characterise 

 families usually accepted as lower, there can be no doubt that the loss 

 of the abdominals 1, 2, and 3, as free segments was a useful 

 advance, and that a segment once becoming fixed never after- 



