GREGORY, PRESENT STATUS OF ORIGIN OF TETRAPODA 355 



dontidse pass almost imperceptibly into the Osteolepidae, which have 

 rhombic scales and usuall}' exhibit the fusion of the bones in the frontal 

 portion of the skull and in the mandible/' Goodrich (1909, p. 285), on 

 the other hand, places the Holoptychiidae and the allied Glyptopomidse 

 first, the Osteolepidae next, the Ehizodontidse third and the Onycho- 

 dontidag last. He says (p. 285) : ''^It is evident from the structure of 

 the paired lins that the Ehizodonts differ considerably from the previous 

 three families [Holoptychiidae, Glyptopomidse, Osteolepidae], and they 

 should perhaps not be included in the same order/' 



The Osteolepid^ present the following assemblage of characters which 

 in view of all that has been said above I must regard as primitive. 



1). Their paired fins are obtusely lobate, i. e., incompletelj^ "archipterygial'' 

 and less different from the median fins, in contrast to the extremely 

 acute "archipterygial" fins of Holoptychiidse. 



2). The scales are rhombic, like those of the earliest Actinopterygii, and they 

 retain the fully developed ganoine and cosmine layers, which are pro- 

 gressively lost in the Rhizodontid?e and highlj^ modified in the Holopty- 

 chiidffi (c/. Goodrich, 1909, pp. 217, 285). 



3). In the Osteolepidae the dermal rays of the median and paired fins are less 

 produced than in the Rbizodontidse. 



4). The caudal fin in Osteolcpis is of a primitive heterocercal type, whereas in 

 RhizodontidiG it externally approaches the homocercal type (through 

 Gyroptychius) . 



5). Osteolepis had a persistent unconstricted notochord without ossified centra, 

 but the larger and more progressive members of both families [Mega- 

 lichthys, Eusthenopteron) had ring centra. 



6). As to the rostrum of Osteolepis representing a "fusion" of elements I have 

 given reasons above (p. 327) for the contrary view that the nasals, 

 dermethmoids and other elements had not yet been divided by sutures. 



In brief Osteolepis appears to offer an ideal stem form for several 

 phyla. One line with progressively cycloidal scales and intermediate 

 characters of the skull-top seems to have led through Gyroptychius (fig- 

 ured by Pander, 1860, Tah. 6, 7) to Tristichopterus (Traquair, 1875, pi. 

 XXXII) and Eusthenopteron (Hussakof, 1912, Fig. 2). Another line, 

 preserving the rhombic scales, perhaps passed through Diplopterus (Pan- 

 der, 1860, Tab. 4) and culminated in Megalichthys (Wellburn, 1900, pi. 

 XIII; cf. our Fig. 7). The skull of Rhizodopsis, as figured by Traquair 

 (1881, Figs. 1, 2) seems to be allied in pattern to that of Megalichthys, 

 as figured by Wellburn (1900, pi. XV) and yet the former is referred 

 to the Ehizodontidae, the latter to the Osteolepid* : and there are otlier 

 indications that these two families are very closely allied. 



In the Holoptychiidse the scales have become large, fully cycloid and 

 deeply overlapping and have the cosmine layer sculptured, but the Glypto- 



