418 ON THE GENUS CORDULEPHYA, 



Although this species is much larger than Cordulephya, its 

 abdomen is scarcely more corduliform than that of C . mon- 

 taiui; and it possesses a colour-pattern of thorax and abdo- 

 men which is remarkably similar to that of Coi'dulephya, the 

 only difference being that the red of II esperocordulia becomes 

 dull orange in Cordulephya. If we consider the extreme 

 peculiarity of the colour-pattern of these two insects, and 

 that it is not even approached by any other known Corduline, 

 we shall see that it constitutes a strong argument in favour 

 of their close relationship. The ancestors of Cordulephya, 

 before they came in contact with the eastern stream of 

 invaders, must have been closely allied to Ilesperocordulia 

 as we find it now. Points of importance, here, in the wing- 

 venation of Ilesperocordulia are (i.) the fact that recession 

 of the hindwing-triangle is not absolutely accomplished; (ii.) 

 the fact that the anal loop has no apical extension; (iii.) the 

 fact that the arculus-sectons are not fused. It is not necessary 

 to suppose that the ancestor of Cordulephya had prog-ressed 

 even so far in a Eurordulia/i direction as H esperocordulia 

 now has. But even if it had, it would not yet have reached 

 a stable position, for variation is still evident in that species. 

 So much easier, therefore, would the start of the reduction- 

 process be. Placing the wings of II esperocordulia berthoudi, 

 Cordulephya inontana and C . pygmcfa in order, side by side, 

 we see 



(i.) Three consecutive stages in the reduction of the tri- 

 angles, by procession from the arculus and the ascent of the 

 upper cross- vein. 



(ii.) Three consecutive stages in the reduction of the anal 

 loop, from ten cells to iowx or three, and then to two strong 

 cells only. 



(iii.) A tendency to strengthen the arculus by the begin- 

 nings of a fusion between the sectors. 



