CLASSIFICATION OF DIPTERA 59 



tween the Streblidae and Borboridae, than between the 

 Borboridae and Empididae, for instance. 



I must insist, notwithstanding Osten Sacken's rather 

 strongly asserted views to the contrary, that there is no 

 distinct limitation between the Nemocera and Brachycera. 

 If the reader will turn to the characters given in the fam- 

 ily table further on, he will, I believe, appreciate the 

 force of my statement. The antennae of Rhachicerus, a 

 'brachycerous' fly, are in some respects among the most 

 primitive or generalized of all diptera: we know of no 

 living genus, unless it be Cerozodia, from which it could 

 have been derived; doubtless some ancient tipulid may 

 have been the forbear of the genus. Of course we can 

 not positively say that these insects may not have revert- 

 ed in the structure of the antennae from a specialized to 

 a generalized condition, but that is very improbable in- 

 deed — evolution is irreversible. It would seem, however, 

 that all the Brachycera, save possibly Rhachicerus, have 

 been evolved from a common ancestor, otherwise we are 

 at a loss to explain the seemingly strange fact that the 

 number eight is so common for the segments of the fla- 

 gellum, and is never exceeded, with the above exception. 

 If the 'posterior cross-vein' of the Brachycera is identi- 

 cal with the vein at the outer end of the discal cell in the 

 Tipulidae and Rhyphidae, then it is apparent that all 

 families, save these, of the Nemocera, are excluded from 

 the ancestral line of the Brachycera. 



L,ameere, indeed, would divide the Nemocera and Bra- 

 chycera between the Mycetophilildae and Bibionidae. 

 But that is only begging the question, the two families 

 are absolutely coalescent. The fact is that the structure 

 of the head, antennae, and palpi changes so gradually 

 from the more generalized to the more specialized forms 

 that nowhere can we draw an impassable line between 

 the groups. 



