is shorter, thicker, more cylindrical, the thoracic plate simple-haired, 

 the tubercles more simple or generalised, i and h more or less 

 separated, iv and v comparatively simple, the extra tubercles behind 

 the spiracles small or absent, whilst strangely enough secondary skin- 

 hairs are very abundant. On the other side, the larva is longer and 

 flatter, the thoracic plate with abundance of hairs, the tubercles form- 

 ing complicated warts, i and ii tending to approximate or join together, 

 iv and v forming a common wart, supernumerary tubercles behind spi- 

 racles present, whilst secondary scattered skin-hairs are few or absent. 

 The Oxyptilids on the one side form, in some of their species, the 

 most difficult exception, in that, as outside feeders, they remain 

 abundantly haired, whilst, on the other side, and, as if suggesting 

 that the presence of secondary skin-hairs is a generalised character, 

 spilodactyla, niveidactyla, and tridaclyla, the little group that I have 

 named Merrifieldiids, are supplied with them, spilodactyla more 

 abundantly than any other on this side of the phylum. 



The plume pupa is a very specialised one, characteristic in many 

 ways, and so distinctly sni generis that once one knows a plume pupa, 

 one can scarcely be mistaken about them. Their mode of suspension, 

 by the wonderful cremaster which, partly on the eighth and partly on 

 the tenth abdominal segments, supports it vertically or horizontally 

 against any flat surface, and, in conjunction with its free segments, 

 although allowing of scarcely any lateral movement, permits it to turn a 

 wonderful somersault movement, arching itself over in such a manner 

 that the head is brought into close contact with the hinder segments 

 if it be disturbed, is very remarkable. There is no other pupa 

 known which has three free segments in the $ and four in the $ , 

 which is unable to progress independently or has its movements fixed 

 by a cremastral attachment. 



But perhaps a still more important item than the generalised 

 characters here noted is the special character noticeable as character- 

 istic of the pupa? on the two sides of the plume phylum. One, the 

 Platyptiliid, is smooth, the primary tubercles usually provided with 

 single setae at their normal positions. In the Alucitid pupa they are 

 hairy, and the tubercular larval warts are carried over as pupal 

 structures, there being quite an abundance of well-developed hairs 

 on many of the pupal tubercles. 



Here, again, is a marvellous exception, for in the Oxyptilids again. 

 we find that, whilst the greater number of pupae are largely Platy- 

 ptiliid and smooth, a few, of which our heterodactyla is the best ex- 

 ample, are very hairy, and, in fact, of all others, this species, super- 

 ficially, comes nearest to pentadactyla. One can only assume that, 

 as they agree in all the other characters with Platyptiliids except in 

 the wart structure of larva and pupa, it has arrived at this mode of 

 specialisation by an independent route. 



To complete the comparison between the Platyptiliid and Alucitid 

 branches one has to notice that, besides the one-spina frenulum in 

 the ? of the former, and the two-spina one, in the ? of the latter, 

 the Platyptiliids have the forewings with an apical and hind angle to 



