GREGORY: FISH SKULLS 179 



habits, but in some respects remains more primitive than the latter. It has no opisthotic, 

 no teeth on the maxilla, an eye-muscle canal closed behind, and an opercular bone very 

 narrow in front; but, on the other hand, it possesses two surmaxillae and an ossified first 

 pharyngobranchial in addition to the spicular. Alepocephalus resembles Gonorhynchus in 

 possessing an epibranchial organ, borne by the fourth and fifth arches, and in possessing a 

 cartilage which may be identified as the fifth epibranchial; but the list of resemblances is 

 soon exhausted. 



"On the other hand, the Salmonidse, though offering no close resemblances to the 

 Gonorhynchidae, consist of a variety of forms but little specialized and highly plastic. 

 For the purposes of comparison the genus Salmo is less suitable than such a form as 

 Coregonus, for the Salmons have an excess of cartilage, presumably of secondary origin, 

 in the cranium, and no membranous interorbital septum such as Coregonus has. It may 

 be pointed out that within the family Salmonidse there are forms, such as Coregonus 

 oxyrhynchus, with prominent snout and reduced mouth with no teeth. 



"Although a study of the cranial osteology of the Gonorhynchidae and Salmonidse 

 cannot bring forward direct evidence of affinity between these families, the hypothesis of 

 the descent of the Gonorhynchidae from the Salmonoid stock is open to little objection of 

 any serious import." 



Notogoneus. — A comparison of Ridewood's figures of the skull of the existing Gonor- 

 hynchus greyi with Smith Woodward's figures (1896) of the extinct Notogoneus osculus from 

 the Eocene of Wyoming and of Notogoneus squamosus from the Upper Eocene of France 

 shows that the ancient form had already acquired the chief characteristics of the gonor- 

 hynchid mouth parts, although slightly less specialized in details. The ascending or 

 coronoid process of the dentary and the coronoid process of the articular-angular are both 

 higher than in the recent type, the snout less elongate, the opercular less reduced, the 

 subopercular large, with four deep clefts on its hinder border. According to Smith Wood- 

 ward (1901, p. 275) and Ridewood {\90Sb, p. 363) there are no pterygoid and lingual teeth 

 in Notogoneus. The skull as a whole seems relatively less elongate and depressed than in 

 the recent genus. Even as far back as the Upper Cretaceous of Mt. Lebanon and West- 

 phalia, according to Woodward (1901, p. 273), the genus Charitosomus much resembled 

 Gonorhynchus not only in general characters but even in the form of the maxilla. A patch 

 of bluntly conical teeth in the throat are found "just above the ceratohyal as if they had 

 been fixed upon the hyoid arch." It seems not impossible that these were really located 

 on the basibranchial as in Gonorhynchus and that the ceratohyal had merely pressed against 

 them, while the opposing set above them appears to correspond to the entopterygoid teeth 

 of Gonorhynchus. Thus the isolation of the specialized family of Gonorhynchidae among 

 the Isospondyli, even in Upper Cretaceous times suggests that their line of ancestry may 

 run back to the Jurassic Leptolepidae, which are truly generalized isospondyls. 



This inference seems to be supported by the characters of the otolith of Gonorhynchus, 

 which as described by Frost (1925, pp. 161, 162) conforms in general to the elopine type, 

 while approaching the mormyroid type in certain features. 



In his great work on the "Classification of Fishes" (1923, p. 120), Jordan makes the 

 following remark: "It is doubtful whether the extinct forms of this group {Notogoneus, 

 Charitostomus) really belong to the same family as the living Gonorhynchus." If this 

 means that Notogoneus is not at least closely related to the direct ancestry of Gonorhynchus, 



