GREGORY: FISH SKULLS 173 



dentary and the quadrate has brought the mouth beneath the orbit instead of far in front 

 of it. 



Still more specialized is the eel-shaped Gymnarchus, which is said to propel itself 

 through the water entirely by the action of its elongate dorsal fin, forward and backward 

 with equal facility (Boulenger, 1904, p. 552). Its strange skull, which has been well figured 

 by Erdl (1847, PI. V) and especially by Hyrtl (1856, Pis. I, II), might well be derived from 

 that of Mormyrops (Fig. 62), which it resembles in the emphasis of the olfactory and otic 

 regions and the retention of strong mandibles, which likewise have an even row of stout 

 teeth; the eye also is small and placed far forward, above and slightly behind the quadrate- 

 articular joint; the lower part of the hyomandibular-quadrate is horizontal. The skull of 

 Gymnarchus is more specialized than that of Mormyrops in being secondarily shorter, in 

 having a large fenestra below, a stout sutural junction of the quadrate' and hyomandibular; 

 also in the almost horizontal position of the preopercular, in the forward displacement 

 of the opercular and other features. In a classification based on skull structure Gym- 

 narchus would be placed following Mormyrops, its eel-like adaptations being undoubtedly 

 secondary. 



The chondrocranium of a forty-third-day larva of Gymnarchus niloticus, as figured by 

 Assheton in the Budgett Memorial \'olume (PI. XX), shows the usual foreshadowing of 

 adult conditions, except that the cranium is far less elongate than in the adult stage. The 

 semicircular canals are very large and thick. In the midst of them is a large "temporal 

 vacuity" on the lateral wall of the cranium, which receives the "anterior bulb" of the 

 air-bladder (PI. XVIII, Fig. 31) much as in Notopterus (cf. Bridge, 1900, PL XXXVII). 



As to the phylogenetic relationships of the families just treated, Ridewood (1904f, 

 pp. 212, 213) writes as follows: 



"On the whole, the study of the craniological characters impels one to the conclusion 

 that the families Mormyridse, Notopteridae and Hyodontidae, though more closely related 

 inter se than is either family with any other family of Malacopterygian fishes, are not more 

 intimately related with one another than was previously assumed to be the case . . . the 

 cranial characters of the families are so conflicting that any phylogenetic arrangement 

 based upon them is out of the question. The three families must remain, as hitherto, 

 the terminals of a radiating system." 



As Boulenger points out (1898, p. 778), the Mormyridae cannot rightly be grouped 

 with the Ostariophysi of Sagemehl (1885, p. 22), as Jordan and Evermann have done 

 (1896, p. 114), since they possess no Weberian ossicles nor other modification of the anterior 

 vertebrae. It is true that Garstang (1932) has recently assumed that the common possession 

 of otic diverticula of the swim-bladder in mormyroids and Ostariophysi is alone sufficient 

 to warrant the bracketing of these two groups along with some others in a superordinal 

 division Otophysi; but the profound differences of mormyroids and characins in skull 

 structure at least lend no support to such a procedure. The absence of a separate sym- 

 plectic in both mormyroids and siluroids is evidently nothing more than a matter of con- 

 vergence; the symplectic is absent also in the eels. The study of the skull of mormyroids 

 shows also that they can have no close affinity with the Esocidae, with which family- 

 Johannes Miiller associated them. According to Boulenger (1898), the nearest allies of 

 the mormyroids are to be found in the Albulidae, as suggested by Valenciennes in 1846. 

 Ridewood then states that he regrets that the study of the skull brings forward no evidence 



