FOSSIL AND RECENT 207 



placed the plethodont Ananogmius with the albulids and expressed doubt as to 

 whether Plethodus, Thryptodus and Pseudothryptodus should be affiliated with the 

 Osteoglossidae or the Albulidae. There are a great many characters suggesting 

 that the plethodonts are in no way related to the albulids, including details of the 

 cranial roof, infraorbitals, pectoral girdle, vertebral column and caudal skeleton. 

 With respect to the dentition, although the teeth of plethodonts are developed on 

 the same bones as in albulids, those that show a definite adaptation towards crushing 

 (Plethodus and Ananogmius) have molariform teeth coalesced into enlarged dental 

 plates with a histology resembling that of the ' tubular dentine ' of dipnoans (Pat- 

 terson 1967c). This is in contrast to the well-defined, discrete teeth of albulids. 

 Together with the many other anatomical differences, especially the trenchant differ- 

 ences in caudal anatomy, this suggests that the albulids are not related to the pletho- 

 donts and that the crushing dentition in these groups is a convergent development. 



The dentition on the parasphenoid of albulids needs support to resist the strong 

 upward bite. Thus it is not surprising to find an ossified interorbital septum in 

 Albula and Dixonina. The albulid interorbital ossification is formed by ventral 

 extension of the orbitosphenoid and the enlarged basisphenoid. The parasphenoid 

 is further braced by sutural contact with the lateral ethmoid. A similar method of 

 bracing the parasphenoid is seen in 0. lewesiensis, which has an equally large para- 

 sphenoid. In this form, however, the interorbital septum is incomplete. It is 

 interesting to note that in some plethodonts (Patterson 1967c : 227) the interorbital 

 septum is also ossified. The sporadic occurrence of an ossified interorbital septum 

 suggests that this character is not a good phylogenetic criterion. 0. latifrons, 

 which has a very broad parasphenoid, and Lebonichthys, in which the parasphenoid 

 is virtually identical with that of Albula, both lack an ossified interorbital septum. 



The branchial arches of albuloid fishes (only completely known in the Recent 

 forms) differ from the primitive elopoid type in lacking a fifth epibranchial and in 

 having the first suprapharyngobranchial represented by cartilage only (the second 

 is absent in albuloids). The dentition associated with the arches is more distinctive. 

 In the Albulidae and Pterothrissidae the basibranchial tooth plate carries a specialized 

 dentition (see above). The dentition on the other gill arch elements consists of a 

 few rounded tooth plates which are clothed with tiny, needle-like teeth. These 

 tooth plates are few in number and totally unlike the elopoid condition, where the 

 tooth plates form a continuous coating on the oral surface of the gill arches. The 

 albuloid gill-rakers are few in number and only developed to any extent on the first 

 branchial arch. Each gill-raker is club-shaped, the stem being narrow while the head 

 bears a cluster of minute teeth similar to those on the tooth plates. 



The dentition associated with the branchial arches of the Osmeroididae is incom- 

 pletely known. The basibranchial tooth plate of 0. levis is very much like that of 

 Elops and is thus primitive. Upper pharyngeal plates of 0. latifrons are rounded, 

 as in albulids and pterothrissids. There is no substantial guide in the known gill 

 arch anatomy to suggest a link between the Osmeroididae and other albuloids. 



Among elopiforms the albuloids show a distinctive hyoid bar, with the upper and 

 lower hypohyals offset, both with respect to one another and to the anterior cerato- 

 hyal. The afferent mandibular artery (which in most lower teleosts runs through 



