Occasional Papers of the Museum of Zoology 7 



and more easily interpreted than figures of any other part it seems 

 worth while to attempt some statement of relationships based on a 

 study of this organ, thus preparing a tentative outline of the 

 classification of the subfamily to be corrected and revised by later 

 studies of other organs. 



Since de Selys' Monographic des Gomphines was published our 

 definition of Gomphinae has grown by reduction, and it may be 

 noticed that three of the five palaeogenic groups defined by Till- 

 yard, in The Biology of Dragonflies, are groups which de Selys 

 included in the Gomphmae. The Gomphinae of recent authors is 

 the "Division Integrilabiees " of de Selys, containing over fifty 

 genera and three hundred and ninety species. 



If the wings of the gomphine Archaeogomphus hamatus and the 

 libelluline Hypothemis are compared certain common characters 

 are obvious: the basal crossing of ac, the four-sided triangles, the 

 undifferentiated cross-veins between Mi_3 and M4, and the unsym- 

 metrical forking of Mj-j and M3. The corduline genus Cordu- 

 lephya has these characters less marked, ac is not so basal and 

 Ml- 2 and M3 are symmetrically forked. If the wings of Gomphus 

 dilatatus, for example, are compared with those of Archaeogomphus 

 it will be seen that in the former ac is very distal, the triangles are 

 t3^ical, the cross-veins between Mi_3 and M4 are reduced and 

 specialized, and the forking of Mi_j and M3 is symmetrical. I 

 believe these four characters, in connection with other venational 

 characters, can be employed as valuable criteria in classifying the 

 Gomphinae. 



It is assumed that the four characters as found in Archaeo- 

 gomphus are primitive at least for the Gomphinae and that they 

 may be employed as criteria for determining the more primitive 

 members of any particular series. The association of these charac- 

 ters in forms generally recognized as primitive in both Gomphinae 

 and Libellulinae gives weight to this assumption. The four-sided 

 triangle and the unspecialized cross-veins between Mi_3 and M4 



