iqo8] 



Classification of the (\irdiiliinac 



275 



fore wings four sided) out in the cold, and, quite apart from the 

 theoretical improbability of more than two co-ordinate groups in 

 any evolutionary series, I think he has not discriminated suffi- 

 ciently as to the value of the different characters to be found in 

 the disposition of principal veins. 



Hence it will not be amiss to present in this paper some further 

 studies on the venation of the group and their bearings on the 

 classification. 



What are the fundaniental venational characters in the Cor- 

 duliinae? Undoubtedly they are those connected with the dift'er- 

 entiation between fore and hind wings. For fore and hind wing 

 were originally alike. In each of the larger subfamilies of Odo- 

 nata may be found one or more weak and dimunitive forms that 

 have fore and hind wings much alike still. I copy Martin's figure 



Fig. 7. The wings of Cordulephya pygmaea Selys (after Martin). 



of the little Australian species, Cordulephya pygmaea Selys, in 

 illustration of this for the Corduliinae. Here the two wings cor- 

 respond with remarkable closeness, and may be compared through 

 out, almost cell for cell. And, as were befitting in a form so 

 generalized, the "triangle" of the forewing has not become 

 triangular. Elsewhere** I have shown how the triangle is formed 

 by the approxiniation of the anterior ends of two crossveins upon 

 a neighboring vein. This process is complete in the hind wing 

 of Cordulephya. 



6. Amer. Nat. Vol. WI. pp. 903-911, 1898, and Proc. U. S. Nat. Mus., 

 Vol. 26, p. 717, 1903. 



