12 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 51 



differentiated from each other in the adults on account of inter- 

 mediate forms. Less and still less grows the clearness of limita- 

 tion as we descend through the series, sections, subsections and 

 superfamilies to the families. Limitations clear a little in the 

 families, but it is not until we get to the subfamilies and tribes that 

 we can. from a study of the adults, begin to draw moderately well 

 marked lines and sel fairly concise limits. A moderate degree of 

 conciseness is possible here only because we arc now concerned with 

 divisions sufficiently low in the taxonomic scale to allow the exclu- 

 sion of refractory and disturbing elements, and if necessary put 

 them alone by themselves. Many subfamilies and tribes are seen to 

 stand out as natural groups of genera. 



At first sight it would appear advisable to ignore the higher divis- 

 ions, and drop at once to the very considerable number of subfam- 

 ilies and tribes necessary to the system outlined. But it evidently 

 serves a better purpose to recognize these higher categories, however 

 much their boundaries may he obscured by connectant forms. They 

 are certainly present, and their existence should not be lost sight of. 

 Therefore they should he retained in any taxonomic system as indi- 

 cating steps in the evolution of these flies. They may be kept some- 

 what in the background, with the caution that they can not be cle 

 and concisely defined until the ontogeny of the intermediate forms is 

 known. 



Man}- genera stand more or less apart and do not fall actually into 

 any of the subfamilies. Very restricted groups of such genera, 

 which may he termed refractory on account of either their complex 

 relationships or their apparent neutrality with reference to the 

 various subfamilies, will best be treated directly as tribes, without 

 reference to any particular subfamily. 



Some few genera will prove to be quite isolated, and yet not enti- 

 tled to subfamily or tribal rank. A final system should aim at the 

 definition of as many well-marked subfamilies and tribes as possible 

 to concisely characterize, and the consequent reduction of the num- 

 ber of these isolated forms. A comprehensive table can thus be 

 preparedj including the subfamilies, the non-referable tribes, and the 

 non-referable genera in one synoptic treatment, which will be con- 

 venient for general use. Separate tables can follow defining the 

 genera within each subfamily and non-referable tribe. No attempt 

 should be made to force refractory genera into any subfamily or 

 tribe where they do not fall naturally, or any tribe into any sub- 

 family where it does not clearly belong, or to antagonize natural 

 affinities in any \\a\, or to combine refractory forms in one heter- 



