760 PROCEEDINGS OF THE NATIONAL MUSEUM. vol. xxvi. 



for critical determination of the course of specialization. For instance, 

 forms with unbraced stigma are not to be derived directh^ from other 

 forms which have the sti^^ma braced; and so for every other useful 

 structure; and so for every stage in the development of each. When 

 the records of the several parts (or of the several organs) agree, the 

 arrangement of forms is simple enough. When they conflict — when 

 one form is specialized in this character and the other in that — we 

 are dealing with diflferent lines of development, and the group is to 

 be divided on the most ancient or fundamental character concerned. 

 When a num])er of characters in disagreement seem of equal impor- 

 tance, with no preponderance of evidence in favor of any one as a 

 basis for a first division, only a tentative arrangement of the groups, 

 subject to change after study of other parts (or organs), is possible. 

 Even when a number of characters are studied and all are in accord, 

 and a small group may be arranged with confidence, evidence from 

 additional parts or organs may show the group to be somewhat less 

 homogeneous than it at first appeared. It is obvious that in a genea- 

 logic study that organ or part is most valuable which possesses the 

 largest number of characters of which one may be sure he knows both 

 primitive form and secondary conditions and characters, which may 

 var}' independently. 



It is probable that every single functional organ exhibits develop- 

 mental features that are characteristic of even the smallest groups, and 

 that the true record of relationships is preserved in ever}" organ if we 

 could but read it. While a classification based upon a single organ 

 is necessarily incomplete, the necessity for the incompleteness arises 

 out of our inal)ility to interpret or even to see the significant features. 

 While a classitication based on one organ is necessarily incomplete, 

 it is not necessarily incorrect. It will at least be self-evident that the 

 classification which must prevail because it expresses the concurrent 

 record of all the parts will l)e hastened by the serious and careful 

 study of each character singly, to determine the facts of its origin, 

 development, and utility, and to trace these facts to their logical and 

 necessary conclusions. 



6. This is only a beginning of what should be done in the study of 

 the venation of the order. The distinctive group characters need to be 

 known, not for whole wings alone, but for every part of the wings. 

 The results worked out in this paper are not specific enough to meet 

 at least three immediate, practical needs. It is frequently necessary 

 to determine fragments of wings: 



((/) In food studies, 



Q}) In the study of unknown nymphs, whose developing wings con- 

 tain the full outline of the venation of the imago. Such wings are 

 often imperfectly preserved and are to be removed only in fragments. 

 Their correct determination makes the most exacting demands on one's 

 knowledge of venation. 



