BY R. J. TILLYARK. GOl 



thus get a beginning of an insight into a natural scheme of 

 classification for the Zi/fjopfera, by means of which we may 

 at last hope to separate fairly accurately, into coordinate 

 groups, that great mass of asthenogenetic forms whose wing- 

 venation has converged so remarkably, from widely different 

 ancestral types, to one single petiolate wing-form. 



In making the above statement, I am aware that I am 

 traversing the received opinions of many profound stiidents 

 of the O(]oii(it(i. If the above natural scheme be carried out, 

 it can scarcely fail to indicate forms such as Hemiphli'hia, 

 Selysionevrd, and a mass of genera in the legion Protoneura, 

 not as highly archaic remnants, but as strongly specialised 

 asthenogenetic or reduction-forms, many of which must be 

 regarded as the most advanced Zygo'pterid types ever formed. 

 A careful study of Tlemiphlehia will, I believe, reveal that 

 this is actually the case, and that the loss of the basal cross- 

 vein in the quadrilateral of the fore-wing is the last and most 

 advanced reduction-stage ever reached by any Zi/r/opfe?'ous 

 dragonfly : instead of being, as has been heretofore main- 

 tained, a remnant of an exceedingly archaic formation. It 

 is time now, at any rate, that students of Odonata should 

 refuse to be content with wing-venational studies only — 

 though these must always play a most prominent part in all 

 satisfactory schemes of classification — and that the evidence 

 afforded by life-history studies, larval gizzards, and labia 

 should be more carefully sought after, so that our knowledge 

 of the OdoiKita may be made more complete and satisfactory. 



Returning to the genus Diphlehia, it may be remarked 

 that, though we must regard all Calopteryf/idce as more or less 

 archaic in some degree, there is strong evidence that Diphlehia 

 is one of the most advanced types in the family. It has 

 evidently progressed far along the zygoterous line of advance, 

 the amount of reduction of the prae-nodal area being very 

 great for a C'alopten/yid . Of the six remaining antenodals, 

 tiro Old// are continuous across the subcostal space. I think, 

 too, that the practical fusion of the subcubital and radio- 



