CLASSIFICATION OF COLEOPTERA. 393 



arity in the structure of the tarsal claws. But the CerophytidaG 

 are very distinct, and their relationship to the Elateridse is much 

 more remote ; so that, if they are to be included in the same 

 group, it is only because they possibly may be a divergent off- 

 shoot from the ancestral stock. 



The Buprestidfe are in many ways a highly specialized family 

 of beetles, and cannot be ancestral to any of those just named. 

 But at the same time, although it is possible, it is very difficult 

 to see how they can be derived from any of those Elateroid 

 families, the differences being so great, and the connecting links 

 nowhere among existing forms to be found. So far I agree with 

 Lameere. But to me it is equally difficult, and for the same 

 reason, to see how they can be derived from the Dascillidae. I 

 look, then, upon the Buprestidae as a very isolated family, whose 

 origin is almost as much of a problem as is that of the Lamelli- 

 corns or Longicorns. The wing-venation in this family is quite 

 characteristic, easily distinguishable from that of other beetles, 

 but not very easily to be derived from that of any other family : 

 (1st) the median vein (M2) has moved forwards towards the 

 radius, and its short recurrent branch comes off at a considerable 

 distance from the apes of the wing; (2nd) the recurrent branch 

 (R^) of the radius (Pij) lies very close to the latter, leaving only 

 a very narrow cell between ; and (3rd) the cubitus appears to 

 have four longitudinal branches, three of them in a position to 

 correspond with the two marked Cu^ in my figure of the 

 Lampyrid wing on p. 125, the fourth homologous with Cug of 

 the same figure. The cell between Cu2 and A^ is generally 

 complete, and then somewhat elliptical in form, with only a 

 single vein continued from it to the margin of the wing, and 

 this vein might be interpreted as A-^ instead of Cu2, if we did 

 not find both often continued up to the margin, while the 

 anterior three branches are at the same time present. The 

 modifications undergone in the fore part of the wing are not 

 unlike those met with in the Lamellicorns, and might easily 

 enough be explained as derived from the Elaterid or Dascillid 

 type. But it is less easy to account for the additional vein in 

 the cubital area. It may possibly have arisen from the conver- 

 sion of a vein originally transverse into a longitudinal one, which 

 then further develops ; it may be due to the bifurcation of an 

 originally single branch, or it may be an inheritance from some 

 ancestor in which the cubitus had four longitudinal branches. 

 If we accept as correct either of the first two alternatives, then 

 the Buprestid wing may without much difficulty be derived from 

 the Elaterid or Dascillid wing ; but if we consider the last of the 

 three as the most probable explanation, we must look for the 

 origin of the BuprestidsB elsewhere. 



The male genital organs have been investigated in only a 

 few species of Buprestidse, and not yet, so far as I know, in any 



ENTOM. — DECEMBER, 1911. 2 G 



