June, 1S09.1 125 



(^Tenthredinidce, Siricidce, aud LydidcB) aud the Petioliventres, em- 

 bracing all other families. Here, again, one of the sub-Orders only 

 (the Sessiliventres) has any real unity in itself, the other is an un- 

 natural congeries of families differing completely both in structure 

 and in habits. 



A still less felicitous attempt is that to found a division on the 

 food of the larva, and make, as Mr, Cameron has done, a group of 

 Phytophagous Hymenoptera. Not only is this division founded on one 

 stage — and that not the final one — in the development of the insect ; 

 but its author finds himself obliged to upset his own system by ad- 

 mitting to his group of Phytophaga species which he owns to be not pliy- 

 tophagous. Indeed, if the aliment of the larva is to be the ground of 

 division, it is hard to see how the ApidcB themselves can be denied a 

 place among Phytophaga ! Tet no one surely could propose to remove 

 these from their place among the Vespidce, Sphegidce, &c., in order to 

 associate them with the Tenthredinidce. This system does not even 

 give us, as the others do, a single homogeneous group. Both Phyto- 

 phaga and Non-phytophaga (?) will have to embrace phytophagous and 

 non-phytophagous families, and the associations of these families 

 resulting from the proposed division will be in the highest degree 

 unnatural. 



The system of classification suggested by Konow differs from all 

 the above in being not dichotomous, but threefold ; and to the 

 objection that a dichotomous division is in itself to be preferred, he 

 would probably answer that Nature, at any rate, does not seem to 

 prefer it. She, if he rightly interprets her system,* has divided the 

 Hymenoptera into three, and not tivo, sub-Orders — no one of which 

 can be regarded as more primitive than, or transitional to, either of 

 the others. Each sub-Order differs from each of the other two in one 

 of the characters on which the division rests, while from both of them 

 taken together it differs in two of these characters, and agrees with 

 them in 07ie (see his Scheme, infra). 



The characters in question have long been used by systematists, 

 but Konow deals with them somewhat differently, and is led accord- 

 ingly to different results. They are, (1) the single or double 

 trochanter, (2) the junction of thorax and abdomen, sessile or petioled, 

 (3) the alar neuration, three cubital cellules or four originally present. 



* In Deutsch. Ent. Zeit., 1890, p. 228, Herr Konow criticises all attempts to range the 

 phepomena of life in " linear series," and propounds a view that the " circle " or rather the less 

 rigid " ellipse " forms the true boundary of all natural groups. In this fact he sees an analogy 

 to the Copernican " Welt system," which, to the more prosaio English niind might seein a little 

 fanciful. But the theory does not depend upon the analogy, and may be maintained apart 

 from it. 



M 



