ORIGIN or THE WINGS. 63 



is extremely delicate, but at the same time very active. 

 The Campodea-\ike larva of Sitaris, having to face the 

 open air, has a harder skin ; necessity takes it into 

 situations where the true Gamjpodea would rapidly 

 shrivel and dry up. 



The larvae of Lepidoptera and of Tenthredinidce live 

 on plants ; activity to them would be useless, and they 

 do not possess it. The larvae of most Hymenoptera — 

 for instance, of the bee, wasp, ant, Gynips, &c. — of 

 Diptera, and of some Coleoptera, live in circumstances 

 which call for even less locomotion, and have relapsed 

 almost into the condition of their far-distant vermiform 

 ancestor. 



Before concluding, I must say a few words about the 

 probable nature and origin of the wings. Whence are 

 they derived, why are there normally two pairs, and 

 why are they attached to the meso- and meta-thorax ? 

 These questions are not less difficult than interesting. 

 It seems to me that the wings of insects originally 

 served for aquatic and respiratory purposes. From 

 the various modes by which respiration is effected 

 among the different groups of aquatic insects, we are 

 justified in concluding that the original insect stock 

 was, like the Gamyodea, a land animal. 



But in aquatic insects there is a tendency to effect 

 the respiration through the delicate membranous 

 covering of more or less foliaceous expansions of the 

 skin. In the larvge of Ghloeon, for instance, which 

 singularly resembles Gampodea, several of the segments 

 are provided with such foliaceous expansions, which, 

 moreover, are in constant agitation, and the muscles of 

 which in several remarkable points resemble those of the 

 true wings. It is true that in Ghloeon the vibration of 

 the so-called branchiae is scarcely, if at all, utilised for 

 the purposes of locomotion. The branchiae are, in fact, 

 placed too far back to act efficiently. The situation, 

 however, of these branchiae differs in different groups, 

 and it seems indeed probable that originally there 

 would be a pair on each segment. In such a case 



