GREGORY: FISH SKULLS 365 



pushed mouth. Thus would be initiated the wide overlap of the ascending processes of 

 the premaxillae above the mesethmoid, the widening of the vomer and the beginning of 

 the median supra-ethmoid fossa. 



Crapatalus. — In Crapatalus of the Leptoscopidse, as figured and described by Starks 

 (1923, p. 286 and PI. Ill), a further step toward Uranoscopus is likewise shown in some 

 features; but this skull (Fig. 244^) is much more primitive than that of Uranoscopus 

 {cf. Fig. 246) in such characters as the following: the broadening of the braincase is more 

 moderate, the eyes do not so greatly constrict the interorbital skull-roof and there is little 

 or no median frontal-ethmoid fossa. Also the ectethmoids, while projecting widely later- 

 ally, are not nearly so much reduced and crowded as in the more advanced genera. Evi- 

 dently also the nasal chamber was less reduced. The back part of the skull is not so much 

 crowded anteroposteriorly; the pterotics project backward in a more normal percomorph 

 way; the sphenotics project forward; the supraorbitals are not interposed between the 

 f rentals and the sphenotics; the large supraoccipital extends forward under the frontals 

 and widely separates the broadened parietals. 



There are also marked reductions, losses and specializations in this skull. The bones 

 in the undried skeleton Starks tells us (1923, p. 280) were very much thickened with carti- 

 lage but in drying became thin and paper-like. The alisphenoid, basisphenoid and the 

 myodome are all absent. The hyomandibular is longer longitudinally than vertically. 

 The symplectic is extraordinarily large. Thus, as Starks notes (1923, p. 265), this fish is 

 not closely related either to Dactylagnus or to Uranoscopus. Nevertheless in his recent 

 paper on the shoulder-girdle of the teleost fishes Starks (1930, p. 226) notes that "In some 

 respects the shoulder girdle of this form [Crapatalus] is intermediate between Uranoscopus 

 and Dactylagnus.^' This is an important piece of evidence for a common origin of the three 

 families. 



Leptoscopus. — A skull of Leptoscopus macropygus (No. 267, Brit. Mus. Nat. Hist.) 

 gives a strong impression of being closely related to Uranoscopus. The skull roof, however, 

 is not roughened and pitted (perhaps it has sunk too far beneath the skin), the preopercular 

 as well as the opercular is much less extended ventrally. The process on the lateral surface 

 of the pterotic division of the hyomandibular turns forward towards the postorbital process 

 but does not reach it. The dentary has fine teeth instead of large ones. The depression 

 of the articular angular beneath the level of the dentary is pronounced. 



Kathetostoma. — This skull (Fig. 245) is closely related to that of Uranoscopus but is 

 even more specialized in its extreme width. 



Uranoscopus. — In this skull (Fig. 246) the eyes are displaced forward and directed 

 partly upward, but as there is a complete absence of the expanded postorbital fenestrae of 

 Astroscopus (Fig. 247), we may safely assume that those eye-muscles which give rise to the 

 electric organ of that genus had not yet in Uranoscopus become greatly hypertrophied or 

 otherwise modified. Nevertheless the general configuration of the entire facial region of 

 Uranoscopus and Kathetostoma is such that the very extraordinary specializations of 

 Astroscopus might readily be derived therefrom. In other words, as in so many other cases, 

 the "habitus" of the ancestor has become the "heritage" of its descendant, which has gone 

 on to acquire a still later habitus. 



Similarly the interocular channel for the reception of the ascending processes of the 

 premaxillae is far less specialized in Uranoscopus than in Astroscopus since it is widely open 

 and not constricted, there being no electric organs to encroach upon it. 



