i(^o8] Classification of the Cordnliume 279 



aa) Anal loop when present of elongate form Ix.unded distally by a curving 

 ^ basal segment of vein Cu2, and traversed lengthwis^e by a 



more or less distinct bisector; two postanal cells m the tore- 

 wing Sub-family CORDULIINAE s. str. ^ ^ , u 

 b) With fore and hind wings closely similar; no anal loop . Cordulephya 

 bbl Hind wings broader than the fore and with well developed anal loop, 

 c) Anal loop not widened at the distal end, nor with an unsym- 

 metrical prolongation of the apex toward the hind margin. 

 Triangle not retracted to the level of the arculus.* 

 d) Triangle 'of the forewing four sided; that of the hind wmg re- 

 tracted to the arculus, and the hind angle of the wmg grea^t^y ^ ^ 



expanded , ' ' ' ' V ',' ' Z ■' j' ' ' ■' ' , 



dd) Triangle of the forewing normal; that of the hmd wing not 

 retracted to the arculus. , , • , , +- 



e) Anal vein extends almost or quite directly to the hmd angle ot 

 the triangle in the forewing, and the subtriangle is elongate 

 and four sided; veins Ml-3 and M4 descend the arculus its ent- 

 tire len<nh, and are fused for an equal distance beyond it m 

 both wings; post-nodal space (space betv^een the nodus and 



the stigma) but half as long as the antenodal • • .Idionyx 



ee) Anal vein reaches the triangle after being twice strongly de- 

 flected around the angles or the subtriangle; subtriangle 

 three sided; postnodal space two thirds as long as the 



antenodal. , . ,. , •, ^u u- 4. r 



f) Veins M3 and M4 parallel at their distal ends: the bisector ot 



the anal loop divides the basal part unequally, some of 



the cells on the side of it next the triangle being again 



divided. ■ \ u t 



ct) With well developed median supplement m both tore _ 



" and hind wing .■•■■. Idomacromia 



ecr) With no median stxpplement developed in either wmg 

 ^'^^ Nesocordulia 



ff) Veins M3 and M4 divergent to the outer wing margin anal 

 loop consisting of but two rows of cells about equally di- 

 vided bv its bisector. 



s) Anal loop short, and lacking a distinct bisector. 



*^ ^ Gomphomacromia 



o-cr) Anal loop longer, and with distinct bisector. 



"''h) Veins M3 and M4 of the forewing strongly divergent at 

 tips- bisector of ' the anal loop (fig. 8, c) weak and 



angulate between the cells Syncordulia 



hh) Veins M3 and M4 of the forev/ing but slightly diver- 

 gent at the vv^ing margin: bisector of the anal loop 

 better beveloped, usually but little angulated: stigma 

 at least three times as long as wide. 



* Except in Neophya which is specialized so independently it will cause no 

 confusion, and in one or two species of Neocordulia, v/hich are truly synthetic 

 forms but in which the anal loop is rather more squarely truncated on the ends 

 than in anv of the others. N. volxemi Sel., in my opinion, agrees m all essential 

 characters 'with typical Neocordulias, and ought not to be removed to Gom- 

 lihomacromia, with which it does not at all agree in such important matters as 

 the form and relations of the triangles the form of the anal loop, the conforma- 

 tion at the arculus, or the venation about the stigma. ,,1 , • , 



The number of cell rows in a given wing area, like that beyond the triangle, 

 when variable, is always determined merely by the presence or absence ot weak 



^^^^N^ophva is remarkable for the contrast between the two wings in point of 

 specialization. The forewing has a triangle of the most primitive sort, while 

 the hind wing exhibits the maximum of broadening of the hind angle together - 

 with complete retraction of the triangle to the arculus. 



