THE CANADIAN ENTOMOLOGIST 37 



Newman's genus, and did not even give it a place in his dichotomic tables, 

 though he must have known of its existence. 



The characters that distinguish the Leptoperlidse, as here defined, (Text. 

 Fig. 4), are the loss of lA in the forewing, leaving only two anal veins, 2A and 

 3A, of which the latter is always forked; the cubitus of the forewing either 

 simple or once forked ; the absence of archedictyon in the anal area of the hind- 

 wing; the presence of the re-entrant angle at the distal end of Cu2; the presence 

 in the hindwing of a fusion of M3+4 with Cui for part at least of their lengths; 

 the possession, as in Eustheniidse and Austroperlida?, of the archaic type of 

 tarsal joints, mandibles, clypeus and labrum, and the widely separated front 

 coxae. In the larva?, there is a unique development of a rosette of gill-filaments 

 around the anus; no other type of gill is present. 



There remain over only a few very reduced forms of Nemouridse, found 

 equally in Australia, Tasmania, New Zealand and Southern Chile, of which 

 the genus Udamocercia End. contains at present the only described species. 

 These are true Nemourids in the widest sense, the imagines having the cerci 

 reduced to one joint, while the same is true for the larva, which also have no. 

 visible gills. 



Owing to a fortunate meeting with Mr. Nathan Banks, of the Museum 

 of Comparative Zoology at Cambridge, Mass., I have recently been able to 

 discuss my plan of classification for the Notogsean Perlaria with him, and to 

 learn from him more details of the morphology of those genera not represented 

 in our Southern fauna than was possible with the limited material at my com- 

 mand. I wish here to express to Mr. Banks my very grateful thanks for a 

 very illuminating discussion which I had with him, in which he clearly set 

 forth the main characters of the various genera of the Northern Hemisphere, 

 and pointed out what he considered the basic errors of accepted classifications. 

 As soon as I had succeeded in convincing him that the Eustheniida? and Austro- 

 perlidae, as defined in this paper, had no close relationship with the Pteronarcidaj 

 proper, the rest became "plain sailing," and we soon arrived at a complete 

 scheme of classification which illustrates the phylogeny of the Order well, and 

 at the same time offers excellent characters for the systematist. 



The first point to be noted is that the old line of evolution which began 

 with the Eustheniidse and Austroperlida?, is carried on by the Nemouridse and 

 Capniidse. Both these families retain the original form of mandibles, clypeus 

 and frons, while they also keep the primitive widely separated front coxae. 

 As regards their wing venation, both can be developed by further specialization 

 from types found within the Leptoperlida?; but the Capniida have progressed 

 a point further than the Nemouridae, in having lost the fork of 3A in the fore- 

 wing. On the other hand, the Capniida? have retained the original many- 

 jointed cerci; while, in the Nemourida?, these processes are reduced, both in 

 larva and imago, to a single joint. Mr. Banks and I quite agreed that the 

 elevation of the groups of Taeniopteryx, Nemotira and Leiictra to full family 

 rank was not justified; and, in this paper, these groups are considered to be 

 only subfamilies of the Nemouridae. 



The second point to note is that the two families Pteronarcidae and Perlidae 

 (this latter including the Perlodidae, which are at most only a subfamily of the 

 Perlida^) form an evolutionary sideline marked by certain high specializations 



