ii.] METAMORPHOSES OF INSECTS. 43 



much more than this, for external circumstances act on 

 the larvse, as well as on the perfect insect : both, therefore, 

 are liable to adaptation. In fact, the modifications which 

 insect larvae undergo may be divided into two kinds 

 developmental, or those which tend to approximation 

 to the mature form ; and adaptational or adaptive ; 

 those which tend to suit them to their own mode 

 of life. 



It is a remarkable fact, that the forms of larvse do not 

 depend on that of the mature insect. In many cases, for 

 instance, very similar larvae produce extremely dissimilar 

 insects. In other cases, similar, or comparatively similar, 

 perfect insects have very dissimilar larvse. Indeed, a 

 classification of insects founded on larvse would be 

 quite different from that founded on the perfect insects. 

 The Hymenoptera, for instance, which, so far as the 

 perfect insects are concerned, form a very homogeneous 

 group, would be divided into two or rather one portion 

 of them, namely, the saw-flies, would be united to the 

 butterflies and moths. Now, why do the larvae of saw- 

 flies differ from those of other Hymenoptera, and re- 

 semble those of butterflies and moths ? It is because 

 their habits differ from those of other Hymenoptera, 

 and they feed on leaves, like ordinary caterpillars. 



In some cases the form changes considerably during 

 the larval state. From this point of view, the trans- 

 formations of the genus Sitaris, which has been carefully 

 investigated by M. Fabre, are peculiarly interesting. 



The genus Sitaris (a small beetle allied to Cantharis, 

 the blister-fly, and to the oil-beetle) is parasitic on a 

 kind of bee (Anthophora) which excavates subterranean 

 galleries, each leading to a cell. The eggs of the sitaris, 



