2 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. I42 



The other approach, and the one used here, constitutes an attempt 

 to broaden the basis for classification by the evaluation, from a phylo- 

 genetic point of view, of hitherto unused or neglected characters. 

 Examples of this type of approach in progress are Svetovidov's 

 studies of the fish brain, Marshall's investigation of the air bladder, 

 and Orton's work on the possible significance of larval characters for 

 teleostean classification. None of these studies will in themselves 

 provide a teleostean phylogeny, but they can contribute data that will 

 bear on such a phylogeny in two ways. First, they can suggest work- 

 ing hypotheses concerning relationships. Second, they can provide 

 general background information concerning the possible usefulness of 

 a particular structure for classification : which parts of a structure 

 are likely to prove constant and which variable; to what extent a 

 feature is correlated with the nature of the environment or with 

 other characters ; and what has been the general sequence of evolu- 

 tionary development in the feature. Once these things are known, a 

 structure can be added with greater confidence to the stock of data 

 bearing on teleostean relationships. 



All the structures to be dealt with in the present paper are osteo- 

 logical. They are (I) the caudal skeleton, (II) certain aspects of 

 pelvic structure, (III) the superficial bones of the snout region (ex- 

 cept the nasal), and (IV) the development of a protrusile upper jaw 

 mechanism. All these structures are followed in greater or lesser de- 

 tail from the clupeiform fishes through to the percoids. The paper 

 concludes with a summary showing how the information presented 

 would seem to bear on the individual lower teleostean "orders" and 

 with a discussion of possible groupings of these "orders." 



For purposes of the present exposition Berg's (1940) ordinal ar- 

 rangement and nomenclature are adopted. 



Certain general statements about the four structural complexes to 

 be dealt with may be placed here advantageously. The first concerns 

 methodology. It has been found by experience that the members 

 within an order that have been considered "primitive" on other 

 grounds are also usually "primitive" with regard to the four features 

 studied here and are hence of special importance for a study of 

 derivations. Since the objective is an understanding of ordinal re- 

 lationships this paper deals primarily with these "primitive" members 

 within orders. Conversely, specializations that appear to have taken 

 place within an order, e.g., the protrusile jaw of the cyprinoids, are 

 usually summarily dismissed, however interesting they may be 

 structurally. 



