96 BRITISH LEPIDOPTERA. 



character. Still; it seems impossible to frame any general description 

 of each, that might not make, say, Centra vimtla — a Lasiocampid, and 

 Eriogaster lanestris — a Notodont. 



The character presented by the pupa-incompleta of always emerging 

 from the cocoon has very few exceptions. Some Nepticulids are 

 recorded as not emerging from the cocoon, even in species that usually 

 do so. The female pupa of the Psychidae does not do so, but this 

 concurs with the circumstance that the moth herself does not do so, 

 except as the last act of her existence. 



The Pterophorids are very exceptional. They rarely have a cocoon 

 to emerge from, and attach themselves by cremastral hooks to a silken 

 pad that is paralleled in a Tineid family with obtect pupae, viz., that 

 consisting of Hypercallia, Anchinia, and their allies. They have pre- 

 served three free segments, either because they have never had occasion 

 to make it desirable to lose them (as has happened with so many 

 butterflies) , but more probably because it enables them to make that 

 remarkable somersault movement backwards, a movement no doubt 

 useful in repelling or frightening enemies. They, as well as the 

 Hypercallias, have cremastral hooks on the 8th abdominal segment, 

 as well as on the usual 10th, giving an extended and solid hold of the 

 silken pad, and affording a special means of meeting the difficulties of 

 the pupal moult. The pupa? of Agdistis are typically Pterophorid in 

 the head sculpturing, in the free segmentation, in the method of attach- 

 ment, and in dehiscence (dorsal head-piece carrying eyes, &c.) they 

 differ in being smooth and very elongated. The pupae of the Ptero- 

 phori, on the other hand, appear always to be short, very broad and 

 blunt forward, and usually to be rough. There is a strong tendency 

 to a longitudinal subdorsal ridge in the line of the trapezoidal 

 tubercles, and this carries either bundles of hairs or great horns of 

 pupal tissue. The former being more common in those species with 

 hairy larvae, the disposition of hairs on the larvae and pupae being 

 much alike ; the horned pupae are more common in those species whose 

 larvae have simple tubercles. 



The Anthrocerid pupa illustrates a question that may be difficult, 

 perhaps, to answer as regards the pupa-obtecta, viz., how far is the 

 delicacy of the welded portions of the pupal covering a persistence of 

 the delicacy of the whole covering of the primary pupa (Eriocrania) , 

 and how far is it acquired ? In Anthrocera (we may take A. filipen- 

 dulae as fairly typical of the family), the 1st abdominal segment is 

 fixed slightly to the wings, but ail the others are certainly free, yet, 

 where covered by the wings (though not adherent to them), not only 

 the 1st, but the 2nd, 3rd, 4th (and even a scrap of the 5th) abdominal 

 segments have a very delicate colourless covering. It also illustrates 

 how, in the lower forms, it is difficult to define what segments are 

 " free." In a Noctuid pupa, the 5th and 6th abdominal segments are 

 f ree — there is no doubt about it — the three incisions involved admit of 

 abundant movement in all directions. Nowhere else is there any 

 trace or suspicion of movement. In Anthrocera, the 4th, 5th, 6th (and 7th 

 in $ ) are unquestionably movable, the 3rd is slightly fused to the 2nd 

 laterally, but dorsally and ventrally motion is provided for, and may 

 be seen in a living pupa ; the 2nd is not even quite solid with the 1st, 

 and some movement occurs here. The pupal skin, generally, of the 

 abdominal segments, though black and solid-looking, is thin and 



