34 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. I42 



Typhlichthys, however, the premaxillaries are not only nonprotractile, 

 but have no downward hook at the tips, tapering laterally to a point 

 as is usual in fishes. The inference would seem to be that the pre- 

 maxillaries of these two genera represent the primitive condition, and 

 that a peculiar protractile mechanism has developed within the 

 cyprinodonts. 



To summarize the foregoing material, it may be said that the 

 following orders usually placed below the Perci formes have basically 

 protrusile upper jaws: Gadiformes (with Macruri formes), Gaster- 

 osteiformes, Lampridi formes, Beryciformes, Zeiformes, Phallostethi- 

 formes, and Pleuronecti formes. The following orders have basically 

 fixed premaxillaries but have one or more members developing a pro- 

 trusile upper jaw: Clupei formes, Cypriniformes, and Cyprinodonti- 

 formes. 



A few concluding notes may be added concerning certain groups 

 with nonprotrusile premaxillaries. It would appear, as already men- 

 tioned, that all the elements necessary for a protrusile jaw mechanism 

 are present in the basal scopeliform fishes; nevertheless, the final step 

 of actual jaw protrusion has apparently not been taken. What has just 

 been said applies equally well to the Percopsi formes. The possibility, 

 which the author, at least, cannot refute, exists for such groups that 

 the lack of a protrusile jaw is here due to secondary loss. 



The upper jaw structure of the Syngnathi formes is very different. 

 Here, in Fistularia and Aiilostomus, at least, the premaxillaries have 

 no pedicels and are bound by a continuous membrane to the anterior 

 heads of both the maxillaries and palatines. Indeed the whole upper 

 jaw structure appears specialized in a direction which is very different 

 from that of the Gasterostei formes and Perci formes. 



DISCUSSION 



It remains to integrate the four structural systems that have been 

 followed in the preceding sections with one another and with the 

 existing classification of modern teleostean fishes. This can perhaps 

 best be done by taking up one by one the lower teleostean orders as 

 given by Berg. 



Clupeiformes. — In a previous paper the author (Gosline, i960) 

 has dealt with the classification of this group. There a major line 

 was drawn between a division Clupei, including Gonorhynchus and 

 the haplomous fishes, and a division Osteoglossi. The nasal structures 

 noted in the present paper would seem to reinforce such a classifica- 

 tion, for the Osteoglossi never have nasal sacs whereas the Clupei 



