170 



therefore be considered as amphipneustic does not seem to have much 

 weight with those who follow this dichotomic arrangement. Perip- 

 neustic larvae occur in Dolichopodidae and vestigial spiracles are pres- 

 ent in some Strayiomyiidae — facts which point, not to the more primi- 

 tive nature of the species possessing these organs but to the persistence 

 of the latter because of their utility in the larval habitat. 



There must be a further division of the tribes in Diptera, but 

 until we have data upon a larger number of species in the order any 

 proposed subdivision, including that of the present paper, must neces- 

 sarily be merely tentative. I give a figure of a hypothetical genealog- 

 ical tree of Orthorrhapha which illustrates my opinion of the re- 

 lations of the various famiUes (Fig. i). There are defects in the 

 scheme that will probably be obvious to many, and it remains for some 

 future worker to improve on the suggestion here given. The sequence 

 of families in this paper is according to the author's ideas, differing 

 somewhat from that in Williston's "Manual", but subject to amend- 

 ment upon discovery of new data. 



CuliciJae 



Ceratopogoiiiilae 

 Orphiiephilidac 



Rhyphidae 



Chiroiiornidae 



Platyiiridae 



Tipulidae 



Fig. 1. Hypothetical genealogical tree of Nematocera, illustrating the grouping 

 in this paper. 



The nervous system as a means of identification may be elimi- 

 nated because of its limited applicability — in preserved material — but 

 we may with little reserve accept as one of the primary indices to 

 affinities the nature of the respiratory system. Insects normally breathe 

 by means of thoracic and abdominal spiracles, and these are found in 

 the adults and in a great majority of the pupae of the Diptera. It is 

 the exception, however, to find them in the larvae, and as a means of 



