776 REPORT— 1900. 



dontidse, Holoptycliiidse, and Ccelacanthidae, but it is only with the first three that 

 we have at present to deal. The Osteolepida3 and Khizodontidae, which appear 

 together in Middle, and die out together in Upper Palaeozoic times, resemble each 

 other very closel}'. In both we have the paired fins, more especially the pectoral, 

 obtusely or subacntely lobate : there are two separate dorsal tins, one anal, and the 

 other caudal, which is usually heterocercal, though in some geaiera it is more or less 

 diphycercal. lu both the teeth are conical and have the same complex structure, 

 the dentine being towards the base thrown into vertical labyrintbic folds, exactly 

 as in the Stegocephalian Labyrinthodonts, and this along with the lung-like 

 development of the double air-bladder in the recent Polypteridsc has given rise to 

 the view that from these forms the Stegocephalia have oi'iginated. The nasal 

 openings must have been on the under surface of the snout, as in the Dipnoi. 



Of these two so closely allied families we must conclude that the Osteolepidae 

 are the more primitive, as in them the scales are acutely rhombic and usually 

 covered with a thick layer of ganoine, while in the Rhizodontidse they are rounded, 

 deeply imbricating, and normally devoid of the ganoine layer, which, however, 

 occasionally recurs on the scales of H/nzodopsis and the fin-rays of Gyroptychius. 



What then of the structure of the paired fins ? Fortunately in the lihizodont 

 genera Tristichojiterus and Eusthenopteron the internal skeleton of the lobe was 

 ossified, and what we see clearly exhibited in the pectoral of some specimens is 

 striking enough. We have a basal piece attached to the shoulder-girdle and 

 followed by a median axis of four ossicles placed end to end. The first of these 

 shows on its postaxial margin a strong projecting process, while to its preaxial 

 side, close to its distal extremity, a small radial piece is obliquely articulated, and a 

 similar one is joined also to the second and third segments of the axis. The 

 arrangement in the ventral fin is essentially similar. 



In fact we have in the Rhizodontidfe a short uniserial 'archipterygium,' and 

 the question is, Has this been formed by the shortening up and degeneration of an 

 originally elongated and biserial one, or on the other hand do we find here a 

 condition in which the stage last referred to has not yet been attained ? This 

 question is inseparable fj-om the next, whether the Ehizodonts or the Holopty- 

 chians form the most advanced type. 



The Holoptychiidae resemble the Rhizodontidse extremely closely in their external 

 head-bones, in their rounded, deeply imbricating scales, and in the form and arrange- 

 ment of their median fins. But the teeth show a more complex and specialised 

 structure than those of the Rhizodontidse ; the simple vertical vascular tubes formed 

 by the repeated folding of the dentine in that family being connected by lateral 

 branches around which the dentine tubules are grouped in such a way as to give rise 

 in transverse sections to a radiating arborescent appearance; hence the term ' den- 

 drodont.' In this respect, then, the Holoptychiidse show an advance on the Rhizo- 

 dontidse — what then of the paired fins.^ While the ventral remains subacutely lobate, 

 as in the previous family, the pectoral has now assumed an elongated acutely lobate 

 shape, with the fin-rays arranged along the two sides of a central scaly axis exactly 

 as in the Dipnoi ; and though the internal skeleton has not yet been seen, yet, 

 judging by analogy, we cannot escape the belief that it was in the form of a 

 complete biserial ' archipterygium.' 



What, then, ia the condition of affairs in the oldest known Dipnoan ? 



The oldest member of this group with whose configuration we are acquainted is 

 I)ipterus,vfh.\ch. likewise appears in the middle of the Devonian period simultaneously 

 with the Osteolepidas, Rhizodontidse, and Holoptychiidse. In external form it 

 closely resembles a Holoptychian, having a heterocercal caudal fin, two similarly 

 placed dorsals, one anal, and cii'cular imbricating scales, which, however, 

 have the exposed part covered with smooth ganoine. But now we have the 

 ventrals as well as the pectorals acutely lobate in shape, and presumably archi- 

 pterygial in structure ; the top of the head is covered with many small plates, there 

 is no longer a dentigerous maxilla, the skull is autostylic, and the palatopterygoids 

 and the mandibular splenial are like those of Ceratodus and bear each a tooth-plate 

 with radiating ridges. 



Now, comparing Dipterus with the recent Ceratodus and Frotopterus, the first 



