Pisces. 1 75 



artificial group, I came to the conclusion that a valuable character 

 existed in the position of the fenestra at the base of the pectoral fin, 

 situated either in the scapula or between tlie latter and the coracoid. 

 Although I had myself pointed out the varial)le position of this 

 fenestra within the limits of a natural family, the Mormyridae,^ I 

 felt justified in assigning to it the importance of a family cliaracter 

 in the higher group Acanthopterygians, the more so as various 

 authors ^ had even regarded it as worthy of defining Sub-orders or 

 Orders. In this, however, I was mistaken ; and, after having reduced 

 its importance to that of a family character, I must now abandon its 

 use for anything higher than generic definition. 



This result is brought about by the study of the ' Sotithern Cross ' 

 collection. The genus Notothenia, the type of the family Notothe- 

 niiclae, is possessed of a pectoral arch similar to that of the Trach- 

 inidae, Callionymidae, and Gadidae, i.e. with the scapular fenestra 

 between the scapular and coracoid bones, as I have ascertained on the 

 type species of the genus, N. coriiceps, Eichards., as well as on all 

 the other species represented in the British Museum. But now, on 

 examining the condition of things in some of the new fishes dis- 

 covered in Robertson Bay, which, so far as external characters are 

 concerned, do not differ materially from Notothenia, I find, to my 

 great surprise, that the fenestra is situated in the scapula, as in the 

 Leptoscopidae. It is perfectly clear, therefore, that the value of the 

 character in question has been over-rated, and its claim to anything 

 higher than generic importance is out of the question. Even this 

 might be contested by some systematists, but the facility with which 

 the point may be ascertained, without injury to the specimen, by 

 lifting up the skin and muscles at the base of the pectoral fin, 

 should encourage the use of a character which is after all of import- 

 ance and may still help in defining family groups other than the 

 very generalised Notothcniidae. These differ from the Trachinidae, 

 Peixophiidae, Farapercididae, Leptoseopidae, and Uranoscopidae in 

 having a single nostril instead of the two possessed by most Teleosts. 

 I have attempted, in the following synopsis, to enumerate and define 

 the genera grouped under the Nototheniidae. The pectoral arch has 

 not been examined in the genera marked with an asterisk ; otherwise 

 all except Trematomus conform to the type to which Notothenia 

 belongs. 



The air-bladder is constantly absent, and every form examined 



' Poiss. du Bassin du Congo, p. 50 (1901). 



2 Cope, Trans. Amer. PhiloB. Soc. (2), xiv., 1871, p. 458. Gill, Proc. Acad. 

 Philad., 1884, p. 170. Jordan and Evermann, Fish. JS. Amer,, iii., p. 2528 (1898). 



