STRUCTURE OF PUPjE. 303 



found three pairs of conspicuous spiracles, occupying 

 the second, third, and fourth rings, and placed on 

 cylindrical tubes. It is worthy of remark, that the 

 caterpillar spins a double cocoon, the outer of a thin, 

 and the inner of a close texture; and when the respi- 

 ratory gills of the larva are cast off with the old 

 skin, the insect knows how to surround itself with an 

 atmosphere of air in the midst of the water where it 

 resides, the inner cocoon being impervious to moisture. 

 How it contrives to renew this air when vitiated is 

 not yet known, but that it derives it from the water 

 is proved by its always dying if removed into the 

 air.* 



We have formerly described the beautiful appara- 

 tus for respiration in the larva? of the gnat family 

 Cidicidce, Latr.);| but this is rivalled by the organs 

 destined for the same purpose in their pupae, which 

 organs are situated in the head instead of the tail. 

 The pupa, in the same way, changes its position in the 

 water, swimming with its head upwards instead of 

 downwards as before. To enable it to maintain this 

 position, the pupa is farther provided with a fin-tail, like 

 a fish, by which it can move itself at pleasure in the 

 water. It no longer, indeed, requires to take food, but 

 air is indispensable lo it; and water being so unstable, 

 were the pupa incapable of swimming, it would seldom 

 escape being drowned. Its respiratory apparatus 

 consists of two tubes, situated behind the head, on 

 what rnay be called the shoulders. They are of a 

 funnel shape, and project very sensibly, though we 

 should not be disposed to compare them, as Reaumur 

 does, to asses' ears. The transparency of this pupa 

 renders it easy to see in it the parts of the gnat. 



An apparently more simple, though no less remark- 

 able, apparatus for respiration, was observed by 

 Reaumur in the pupa of a small crane fly (Tipida, 



" De Geer, Mem, i, 531. f See page 156. 



