556 Mr. C. T. Regan on the 



was in most respects satisfactory. Several authors have 

 given family names to some of Giinther's groups *, and of 

 these the Eigenmanns t have utilized differences in the 

 structure of the vertebral column and the air-bladder in 

 order to characterize the South-American families. Bou- 

 lenger % separated off as two families the Siluroids with all 

 the ribs sessile, retaining the remainder in one family, 

 SiluridaBj divided into eight subfamilies characterized by the 

 length of the dorsal and anal fins, the freedom or union with 

 the isthmus of the gill-membranes^ &c.^ the majority of these 

 groups being unnatural. 



Bridge and Haddon's § great memoir on the Siluroid air- 

 bladder is of considerable service in working out the classifi- 

 cation. Comparatively little has been written on the 

 osteology of the Siluroids, although Koschkaroff|| has 

 recently given a useful account, with figures, of the skeletons 

 of several genera. The scheme here put forward is based on 

 the examination of a large series of skeletons, and a far 

 greater number of families is recognized and defined than in 

 any previous system. The majority of these are much 

 better characterized than most of the families of Percoids ; 

 the difl'erences between the Lutianidse and Scisenidse, for 

 example, are trifling compared with those between the Syno- 

 dontidse and Doradidse, which have usually been associated 

 in the same subfamily. 



In the following account I place first the Diplomystidse, 

 more generalized than any of the others in the normally 

 formed toothed maxillary and in the simple attachment of 

 the fifth vertebra to the complex. Next the Ariidae and 

 Doradidse are considered, generalized in form and in fin- 

 structure, but aberrant in the loss of the mesocoracoid, and 

 they are followed by the Plotosidfe and Siluridse, which 

 have a very long anal fin, but are primitive in some other 

 characters, such as the many-rayed pelvic fins. Then come 

 the Bagridae, widely distributed in Asia and Africa, and they 

 are followed by the North-American Amiuridse and by a 

 number of Old- World families which may be regarded as 



* Gill (Sinithson. Misc. Coll. xi. 1872) named, but defined only by 

 referem-es to Giinther's Catalogue, the families Hypophtbalmidfe, 'Iricho- 

 niycteridte, Siluridie, Chacidas, Plotoddse, Clariidse, Uallichtbyidas, Ar^i- 

 idte, Loricariidse, Sisoridpe, and Asprediuidfe. 



t Occ. Papers Calif. Acad. i. (ISUO). 



X Cambridge Nat. Hist. Fish. p. 586 (1904). 



§ Phil. Trans, clxxxiv. (B) 1893, p. 65. 



I! Bull. Soc. Nat. Moscow, 1905, p. 209. The genera described are 

 Loricaria, Synodontis, Macro7ies or a related genus wrouglv named 

 AAysis, Clarias, Silurus, Arius, Eutropiv^, Malopterurus. 



