BY R. J. TILLYARD. 725 



army of Zygoplcra. The effect of reduction of the wing-venation 

 has been to produce a superficial convergence in all these forms, 

 mainly by the return of the triangle to quadrilateral form (though 

 not necessarily from an actual three-sided triangle, which was pro- 

 bably never formed with stability), and by the reduction of the 

 anal area of the hind-wing to a comparatively few cells, whose 

 arrangement is then hound to show some similarity in all these 

 forms. But they all retain clear indications of the stock from which 

 they branched, e.g., Ilypothemis, Tetrathemis, and Nannophlehia 

 are essentially Libellulince, Cordulephya essentially CorduUne, 

 Agriogomphus essentially Gomphine. These forms, while based 

 upon ancestral characters that are truly archaic compared with 

 those of the more successful menogenetic forms most closely allied 

 to them, but not necessarily archaic in comparison with other 

 menogenetic forms not closely allied {e.g., Chlorogomphus) , are 

 usually quite highly specialised in some other directions, e.g., in 

 flight, larval development, tibial armature, etc. 



If this theory be correct, we are still a long way off from dis- 

 covering the pomt of origin of the Libelluline from the CorduUne 

 stock. We must, perhaps, confess that this is lost in antiquity. 

 Nevertheless, we may still hope for light on this problem from the 

 discovery of the larva of Chlorogomphus, which may very possibly 

 show characters common to both of the groujis in question. 



One other point is worthy of mention. In the develoj^ment of 

 his theory of the classification of the Libellulina', Dr. Ris jDlaces 

 together in Group ii. the narrow-winged forms, Agrionoptera and 

 Lathrecista, with the normal-winged Orthetrum. We want, there- 

 fore, the larva of one of these narrow-winged genera. If it turns 

 out to be similar to Orthetrum (a very distinct type), it will vindi- 

 cate his classification. If, however, it does not, it will be a serious 

 blow to the evolutionary view expressed for Group ii., and will, I 

 take it, require a reconsideration of the Double-Origin Theory of 

 the Anisopterid anal loops. In my view, forms like Agrionoptera 

 are not, in any sense, ancestral to Orthetrum, but are an astheno- 

 genetic development on a line of their own, converging towards 

 Orthetrum (if one may say so, though the convergence has not 



