THE PHYLOGENY OF THE LEPIDOPTEROUS PUPA. 97 



flexible, and maintains its form largely by being filled witb the living 

 contents. It has, therefore, in these respects, advanced little from the 

 primitive form, which it also exhibits in the way in which the wing- 

 cases, the leg-cases, and the antenna-coverings separate from each 

 other and curl up individually after the moth emerges. It is chiefly 

 in secondary matters that it shows any advance. The eye-covers 

 remain with the face-piece on dehiscence, there are no (or a mere) 

 trace of maxillary palpi, the rows of dorsal hooks are well-developed. 

 It presents another interesting feature, of which the explanation offered 

 may not be correct, but it is one that in any case invites study. In 

 front of the hooks is a piece of delicate membrane, between them and 

 the narrow line forming the anterior border of the segment. This is 

 collapsible, though it does not curl over like the intersegmental mem- 

 brane. In the well-developed pupa-incompleta, that is fully evolved 

 as regards hooks, there is a row at the anterior, and another at the 

 posterior, border of the segment. Our British species of Tischeria show 

 how, within the limits of a single genus, the important stages of the 

 development of this final result may be presented. In one (T. marg- 

 ined) we have the ordinary chitinous points developed into small spines 

 uniformly over the segment ; in another (T. angusticolelld) we see 

 these segregating into a central patch of spines behind the trapezoidal 

 tubercles; in a third (T. complanella) they are massing towards the 

 anterior margin of the segment ; whilst in a fourth (T. dodonaea) they 

 are beginning to form a posterior row of spines. It is rare, however, to 

 find such an illustration as this, and there is usually considerable 

 uniformity within the same genus, or even family. The lower the 

 stage of development, however, the greater variety there usually is. 

 In Anthrocera, there is no great variation, the previous stage must be 

 looked for in some other family. We find it, probably, in Nepticida 

 (not that I suggest Nepticida as an ancestor of Anthrocera, but that the 

 ancestor of Anthrocera probably resembled Nepticida in this particular). 

 Nepticida has a patch of spines in the centre, or towards the anterior 

 margin of the segment. I suggest that, in Anthrocera, this became an 

 anterior row by the failure of the anterior portion of the segmental 

 chitin, now represented by the delicate membrane alluded to. It is not, 

 then, a truly anterior row of hooks as in Tortrtx or Tinea, but a central 

 row attaining an anterior position by means of this special device. 



The Psychides present an assemblage of characters that bind them 

 together as one superfamily, yet with variations that show certain 

 lower groups to be very close to the Adelids, whilst the higher ones 

 have made some approaches to Obtect structure in various subsidiary 

 characters. They are all characterised by having the first two abdo- 

 minal segments fixed, the 3rd free dorsally, the 4th, 5th, 6th, and 7th 

 in the male free. The lower families have a pair of dorsal hooks on 

 the 10th abdominal segment ; the higher have similar hooks, but 

 ventrally, and also a set of dorsal hooks reversed and on the interseg- 

 mental membrane of certain abdominal segments. These posterior 

 hooks, on the intersegmental membrane, are directed forwards when 

 the membrane is extended, and are, therefore, useful for backward 

 movement of the pupa, but can be thrown out of action by the infold- 

 ing of the membrane, when they disappear. To do this without injury 

 to the opposite surface they lie very flat. The very long needles they 

 form on a large pupa, lying closely appressed to the surface, are very 



