276 Annals Entomological Society of America [Vol.1, 



Obviously differences in the approximation of these cross- 

 veins are differences of degree only. Such differences are not 

 fundamental enough for primary divisions of a subfamily. It 

 would be hard to find three genera with less affinity than the 

 three having four sided triangles in this subfamily. Williamson 

 has indicated some diametrically opposite developments in 

 Neophya and Cordulephya, and I have shown^ that the four sided 

 triangle of Pentathemis is not primitive, but secondarily derived 

 from a three-sided one, and is an extreme case of specialization. 



With the exception of Cordulephya, all Corduline genera have 

 fore and hind wings of very different form. This diff'erentiation 

 has been brought about by a number of minor shifts of parts, 

 which I have pointed out in the paper last cited, and chiefly 

 by the broadening of the hind angle of the hind wing and the 

 development of an anal loop for its support. This development 

 has followed two distinct methods: 



I. An anal loop of compact Cordulegasterine form (fig. 8, 

 d, I), externally delimited by a stout straight basal segment of the 

 vein Cu2, has interposed itself squarely between the triangle and 

 the hind angle; and in consequence there is slight tendency for the 

 triangle to recede to the level of the arculus. This is the group 

 MACROMIINAE, which I characterized in 1903. It is William- 

 son's group V with Macromidia added. It ought to be recognizable 

 by the characters I stated even though its constituent genera were 

 not all named. In my judgment it is co-ordinate with all the 

 other Corduliinae s. str. put together. If any one think it insuffi- 

 ciently defined, let him distinguish between the long recognized 

 Libellulinae and Corduliinae s. str. with equal definiteness. 



II. Anal loop tending from the first to be elongate and nar- 

 row and to extend itself outward along the cubital vein, with 

 concurrent recession of the triangle to the level of the arculus 

 (fig. 8, a). This is the group CORDULIINAE s. str. 



These tendencies are least marked in the aberrant genera 

 Gomphomacromia (fig. 8, 6) and Idyonyx, in which the anal loop 

 has remained short but even in these the form of the anal loop and 

 its relations to the basal portion of vein Cu2, are distinct from the 

 Macromian type: In Idionyx and in all except Gomphomacro- 

 mia the bisector of the loop, dividing its two parallel rows of cells 



7. Proc. U. S. Nat. Mus. Vol. 26, p. 718, footnote. 



