722 A NEW SPECIKS OF SAS NOl'ULEDIA, 



rangements of dorsal and lateral spines show nothing in common ; 

 but I do not regard these as beyond specific value.] 



B. Imagines: — 



(1) Colour-scheme. — Somewhat similar in broad outline; very 

 distinct in detail. 



(2) General build and habits. — These show considerable simi- 

 larity, but cannot be held to indicate affinity, as the results of 

 asthenogenetic convergence must necessarily tend to a general re- 

 semblance in form and habits. 



(3) Wing-venation. — Beyond the fact that both are of the nar- 

 row-winged type, there is very little similarity in detail. Contrast 

 a, the Corduline angulated hind-wing of Cordidephya ^ with the 

 rounded Libelluline hind-wing of Nannophlehia ^; b, the differ- 

 ence in position of second antenodal compared with arculus; c, the 

 number of antenodals; d, the points of origin and amounts of 

 fusion of the sectors of the arculus; e, the positions of the cubital 

 cross-veins; /, the points of origin of the superior sector of the 

 triangle; g, the position of the hind-wing triangle (compare also 

 Cordulephya with Tetrathemis or Hypothemis, which are closer 

 to it in this character) ; h, the anal loops (Cordulephya reduced to 

 two strongly marked large cells in C. pygmaea) 



If these two wing-forms were really close approximations to an 

 archaic wing-type, I do not think we could expect so much diver- 

 sity of detail. Assuming, however, that the narrowing of the 

 wing is due to asthenogenesis, the marvel would be, not that they 

 are so close as they are, but that elimination and reduction should 

 fail to produce something even closer (consider, for example, the 

 long but excessively similar set of reduction-forms in the Agrio- 

 nidce). 



(4) Tibial keel.- — A recognised Corduline character. Present in 

 Cordulephya; no sign of it in Nannophlebia. 



The main weight of the above evidence seems to me, however 

 much one may regret the fact, to be strongly against the assump- 

 tion of a close relationship between Nannophlebia and Cordule- 

 phya. It points rather to the fact that the ancestral LibellulirKe 



