SCIENTIFIC NOTKS AND OBSERVATIONS. 287 



we may make Cotism the type, including Tortricids, Sesiids and 

 sundry Tineids, and smaller sections. These all have the first three 

 abdominal segments fused into the thoracic mass, but opening on 

 dehiscence, and have the three following segments [(4, o, B (and 7 in <y )] 

 free. The specimen differs from Cossns, and agrees more with Tortrix 

 in having no beak, which in Cossiis, Semi, Sec, is always more or less 

 developed in the inter-antennal and pre-oral region. It also differs in 

 showing three (nearly four) joints of the tarsus of the 3rd pair of legs 

 projecting free, beyond the wing-cases ; in Cossus, two or three joints 

 are visible, but rather between the slightly separated wing-cases than 

 beyond them. In Z. pyrina, one or two joints so project ; in Senia, 

 also, two joints project. A very decided difference from many of them 

 is in the well-developed maxilla (proboscis), which here extends as far 

 as the end of the wing-cases. In Cossus, this is extremely short ; in 

 Sesia, it extends about half-way in some species, but reaches the ex- 

 tremity of the wing-cases in others. In Sesia, Cossus and Torlrix, the 

 dorsal head-piece is represented by a short narrow plate ; in Z.pyrinaii 

 is not represented externally, though in dehiscence it carries the eye- 

 covers, as in the other forms. This would seem to be the case in 

 Castnia ; no external indication of this head-piece can be detected, but, 

 though dehiscence has not gone so far in the specimen before me as to 

 make this certain, there can be little doul^t that the dehiscence is as in 

 Z. pyrina. The eye-collar (maxillary palp) is developed much as in 

 Cossus, Sesia, &c. The femur shows between the maxilla and the first leg, 

 but is narrow as in Sesia, and not broad as in Cossus, a result in that form 

 no doubt of the stoutness of the maxilla. The anal armature consists 

 of two forward tubercles, as in Cossus, that look like a persistence of 

 the anal prolegs, and of two other tubercles a little further outward and 

 Jjackward that exist as points in Cossus. The variation in the anal 

 armature in Sesia is, however, much greater than this, some species 

 having it rounded and nearly smooth, while others (as bembeciformis), 

 have a nearly complete circle of points. In the armature of the 

 dorsum of each segment, it agrees with all these groups in having a 

 larger anterior set, and a slighter posterior one on the 2nd and (jth 

 abdominal segments, and the anterior set also existing in 7, 8 and 9. 

 In Cossus, the same armament is quite distinct, though small on 

 the first abdominal segment. In Castnia, it is very poorly develoiJcd 

 even in the 2nd, otherwise the arrangement is the same to considerable 

 detail, the forms of the individual spines being very similar. The 

 S2)iracles are much alike, even to the marks of the circumspiracular 

 tubercles, and both present marked impressions of the ventral })rolegs. 

 A dorsal ridge, indicating the line of dehiscence, is very max'ked in the 

 pro-thorax, but not at all in the meso-thorax. This ridge exists in 

 many butterflies. It is very marked in Galleria mellonella, and 

 apparently exists in many distinct families. So far as the pupa goes 

 therefore, Castnia is very close to Cossus and Sesia, and so far as its 

 relations to butterflies is coneerned, it is clearly not in the direct line of 

 butterfly descent, since the dorsal head-piece is reduced to evanescence, 

 j^et still exists markedly in the Hesperids, whilst the other " micro " 

 characters, that it still possesses so fully, hardly exist in the Hesperids. 

 If it be at all then a branch of the main stem which carries tlie butter- 

 flies at its summit, it has diverged to some little distance from, even if 

 it be, perhaps, a later variation than is Sesia. 



