264 



TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY 



longation of the preopercular and interopercular. Probably it Is conditioned in some way 

 by changes in the cerato- and epi-hyah 



In the Atherinidae the skull and jaws are less specialized, the mouth varies from small 

 to moderate size, the teeth are usually small. In the long-bodied Chirostoma diazi (Jordan 



sphot : 



ph 



Basil ichthys bonariensis 



Fig. 139. Odontesthes " Basilichthys " bonariensis Cuv. Kindness of Dr. Tomas L. Marini and the Museo Nacional de Historia 



Natural of Buenos Aires. 



and Hubbs, 1919, PL VIII, Fig. 32) the whole body, head (Fig. 140^) and jaws suggest 

 that the Sphyraenidae are merely giant atherinids. Other atherinids, e.g., Odontesthes 

 perugle {cf. Jordan and Hubbs, 1919, PL III, Fig. 11) and Chirostoma chapale {op. cit., PL 

 VII, Fig. 28) suggest the half-beaks of the order Synentognathi; but Tate Regan (1910a, 

 p. 8) remarks: "The supposed relationship of the Scombresocidae to the Atherinidae is 

 based on a number of resemblances which do not, in my opinion, indicate affinity. The 

 Percesoces are more specialized than the Scombresocidae in that spinous fin-rays are de- 

 veloped, and the features in which the Atherinidae approximate to the Scombresocidae 

 appear to have been evolved within the order Percesoces rather than to be those of the 

 prototype of the group." 



In another direction the Sphymna type is rather closely suggested by the Australian 

 fish Dinolestes lezvini (Fig. 140C), but the very thorough analysis by Starks (1899a) has 

 shown that the latter is a true percoid of the family Cheilodipteridae (Apogonidae) and that 

 it differs from Sphyrcsna in many important characters, including the structure of the teeth, 

 which in Sphymna are set in sockets, in Dinolestes ankylosed to the bone as in the percoids. 



It therefore seems highly probable that the predaceous skull of Sphymna (Fig. 141) 

 is not the most primitive one of the suborder, that its general resemblances to the Dinolestes 

 percoid type is partly an expression of parallelism, that it has been derived from a small, 

 short-jawed type more like that of the atherinids. Nevertheless I am loath to give up the 

 idea suggested by Jordan and Hubbs (1919, pp. 6-8) that the Atherinidae (and with them 

 the Percesoces as a whole) have been derived from true percoids, the ancestors of the 

 Apogonidae and Ambassidae, which appear to afford a suitable source for the cycloid scales 

 of atherinids, for the curiously persistent spinous dorsal of the whole series and for the 

 percomorph features of the skull. It is perhaps for such reasons that Tate Regan (1929, 



