NO. 3 TELEOSTEAN FISHES — GOSLINE I9 



it represents, when present, the holdover of a primitive teleostean 

 (holostean?) feature. 



Under the circumstances it only remains to point out the high de- 

 gree of development of this splint in two lower teleostean groups — 

 the Notacanthi formes and Percopsi formes. Among the soft-rayed 

 halosauriform fishes as represented by Halosaiiropsis, this outer splint 

 is present in fairly typical elopoid form. In the spinous Notacanthus, 

 however, this same structure has been transformed into one of the 

 several pungent spines at the outside of each pelvic fin. The other 

 spines, as indicated by their forked bases, have formed by the fusion 

 of two halves of a segmented ray. (In this they are similar in con- 

 struction to the single pelvic spine of the berycoids and percoids.) 



In the Percopsi formes (fig. 5C) the outermost ray structure of the 

 pelvic fin is about as in the elopoids. This, like so many other fea- 

 tures, indicates the low level of organization of the group. 



At this point the question may well be asked: What is a pelvic 

 spine? If Percopsis has a pelvic spine, then so have Tarpon, Salmo, 

 and others. If the outer splintlike structure should not be considered 

 a spine, then in counting the pelvic spines of Notacanthus should one 

 count the number of pungent elements and subtract one? The diffi- 

 culty cannot satisfactorily be resolved by fiat, but in practice the 

 systematist can easily handle the problem by stating how he is making 

 his pelvic fin counts. As a guide in this matter it may be said that, 

 with the exception of the notacanthids, the author has never found a 

 separate splint developed in those forms that have a true pelvic spine, 

 e.g., holocentrids and percoids. 



In living holosteans and in many lower teleosts there are either 

 three or four radials between the pelvic rays and the pelvic girdle on 

 each side (Sewertzoff, 1934). In Lepisosteus, Amia, and most iso- 

 spondylous fishes (fig. 6A) the innermost radial (the metapterygium 

 according to Sewertzoff) is enlarged and runs partly under and partly 

 along the inside of the innermost ray. Lateral to this innermost 

 radial there are either two or three smaller nodules of bone that are 

 more or less hidden between the two halves of the ray bases. These 

 nodules may be, and perhaps always are, enclosed in cartilage. 



To summarize concerning these radial elements as a group, the 

 author has found them well developed in isospondylous fishes, in 

 Aulopus among iniomous fishes, and in Holocentrus among the bery- 

 coids. However, a number of lower (and apparently all higher) 

 teleosts lack separate radial ossifications. Indeed, even among the 

 haplomous fishes the pelvic rays seem to articulate, at least in part, 

 with a cartilaginous area that contains no radial ossifications. 



