454 ON THE GENUS NANNODYTHEMIS, 



no doubt that it is so subject to variation that it is a dangerous 

 guide to generic division. In the case of Nannodythemis as 

 defined by Brauer, the importance attached to it has been so 

 great that, if insisted on, we are driven to the absurdity of 

 placing the male of N. Dalei in one genus and the female in 

 another. This is sufficient to show that it has no generic value 

 whatever. But rightly regarded, it possesses for us a far deeper 

 significance. For these "quadrilateral" Libellulince are survivals 

 which take us back to a period long before the now dominant 

 genera Diplacodes, Orthetrum, Libellula and many others had 

 been formed. As Nannodythemis , Nannophya, and Tetrathemis 

 are to the dominant Libellulinw, so are Neophya, Cordulephya 

 and Pentathemis to our present-day Corduliince (a group that 

 can scarcely be called dominant). And in these three closely 

 allied species of Nannodythemis we see taking place before our 

 very eyes that excessive variation in the region of the triangle 

 which was probably a heritage of an earlier period, when the 

 " triangle " of the Anisoptera first became differentiated from the 

 simpler quadrilateral cell of the Zygoptera. N. australis repre- 

 sents probably the oldest form, and tracing the gradual decrease 

 of " abnormality " in the hindwing triangle through N. Dalei to 

 N. occidentalism we note the concurrence of greater size and more 

 powerful build; the latter species suggesting at once that it would 

 take but another step, viz., the change from an abnormal fore- 

 wing triangle to a normal one, with another corresponding 

 increase in size and strength of build, to give us the true Austra- 

 lian type of Diplacodes as represented by our D. haematodes or 

 D. bipnnctata. Nannodythemis then points to us the way by 

 which the great dominant group of present-day Libelhdince have 

 ascended in the scale of development, and it is not impossible 

 that a careful study of this and allied genera, both in the 

 Libellulince. and Cordidiinos, may yet reveal the exact hidden 

 homology between those portions of the Anisopterid and 

 Zygopterid wings which lie close to the arculus. With the aid 

 of more material and careful study of individual variations in 

 each species, I hope to give later on in another paper some 



