340 CLASSIFICATION OF FISHES. 



not enter among these types. The two primary divisions 

 of this sub-family may be thus characterised : — In the 

 first, — to which we retain the name of Mystus, as 

 originally proposed by Artedi and all the old ichthy- 

 ologists^ — we see the general form of Pimelodus, united 

 with some of the characters of Loricaria : the parts 

 before the dorsal fin, for instance, are protected by bony 

 plates ; and on each side of the body is a series of 

 smaller ones, placed in a single row, so as to form a 

 lateral line, armed with prickles, as in our M. costatus 

 (Bl. 376.). This structure, in fact, is precisely analogous 

 to the plates of the genus TracMnus among the Scom- 

 beridce, and many other instances of spiny or raised 

 lateral lines to be found in analogous groups. The simple 

 plates upon the head are not, however, confined to this 

 genus ; nor can they be looked upon as a primary generic 

 character ; since they occur in most of the Indian 

 Pimelodinee, and even in some of the Silurince. In the 

 next or most typical genus, Pimelodus, the sides of the 

 body are always naked : the species of this group are 

 exceedingly numerous in the rivers of India, and several 

 are found in those of Tropical America. We have 

 failed, however, in discovering such characters as would 

 separate these species geographically : for although the 

 majority, if not all, of those with the head entirely 

 naked seem restricted to America, yet those in which 

 this part and the nape are covered with a buckler, are 

 found, with little or no essential variation of structure, 

 in both hemispheres. Among the Indian species, minor 

 differences will be observed, sufficient to constitute sub- 

 genera, easily recognised, and therefore much better 

 understood than if we attempted to arrange them by their 

 teeth, — an attempt which has so signally failed in the two 

 " sub-genera" just alluded to, viz. Pimelodus and Bagrus. 

 In some of those Indian species which we suspect are 

 typical of the genus Pimelodus (as we now propose to re- 

 strict it), we observe that the adipose dorsal fin is of such 

 considerable length as almost to fill up the interval be- 



