720 A NEW SPECIES OP NANNOPHLEBIA 



(3) The average length of the legs in Corduline nymphs (and 

 imagines also) is considerable greater than in the LibelUdince. 

 Nannophlehia may be said to have rather long legs for a Libellu- 

 line, but distinctly shorter than we might expect for a Corduline 

 (compare C ordulepliya) . This character may, therefore, be held to 

 agree with the conclusion stated just above. 



(4) In Corduline nymphs there appears to be a much gi-eater 

 tendency to the retraction of segment 10 of the abdomen, and the 

 enveloping of it by segment 9, (or sometimes even 9 and 10 may be 

 retracted into 8). There are, however, many exceptions to this. 

 Also, the appendages in Corduline nymphs are usually somewhat 

 longer than in the Libelluhnce, though the difference in some par- 

 ticular cases is not very great. In so far as these characters may 

 be judged to have any value, Nannophlehia must be held to be dis- 

 tinctly Libelluline. 



(5) The general outline of the nymphs of the two groups is 

 somewhat varia])le, but there is a type that is common to both, 

 viz., the smooth oval body, triangular front, and rounded post- 

 ocular lobes of the head (compare Diplacodes, Hemicordulia, C or- 

 dulepliya) ; and to this general type, Nannophlehia belongs. 



Reviewing tlie above evidence, I am inclined to conclude as fol- 

 lows : — 



Group i. of the Libellulinre., as proposed by Dr. Ris, is justified 

 as an archaic group, probably heterogeneous, but containing all 

 those asthenogenetic remnants that still exist, of the earliest off- 

 shoots in a Libelluline direction from the 'pa,rent{Corduline)stera. 



We have now to face a further and most interesting problem, 

 which may be stated as follows. Both the Libellulinm and the 

 Corduliince possess archaic forms which appear to be fairly closely 

 allied; e.g., Group i. of the Libellulina', Cordulephya in the Cor- 

 duliince How far is this resemblance due to real affinity; how 

 far is it due to asthenogenetic convergence? The answer to this 

 contains within it the solution of the vexed problem of the genesis 

 of the Libelluline anal loop. If we can prove that Nannophlebia 

 and Cordulephya, for instance, are really close allies, and not 

 merely alike by convergence, we have a strong ground from which 



