48 



SCIENCE 



[N. S. Vol. XXVI. Ko. 654 



unrelated dental plates. The chances are in- 

 finitely small that in the fossil in question 

 two such plates, if once separated, could have 

 accidentally come to lie in such accurate ap- 

 position. And until more perfect material is 

 forthcoming, the present specimen remains of 

 paramount value, none the less so since, as 

 the writer pointed out, the faceted surface of 

 these combined plates corresponds to the in- 

 dented area of the " mandibular," which is 

 present in the same fossil and must have 

 apposed them. By this view, also, the denti- 

 tion of one arthrodire can best be explained 

 in terms of another, the smaller, more irregu- 

 lar " premaxillary " of Mylostoma becoming 

 the homologue of the smaller and more irregu- 

 lar " premaxillary " of Dinichthys, and the 

 longer oblong " maxillary " to the long " shear 

 tooth " of the latter form. It is not necessary, 

 therefore, to go afield and postulate a closer 

 affinity of the Devonian arthrodire Mylostoma 

 to a recent lung-fish when a comparison can 

 readily be made with a contemporary form 

 (Dinichthys) , to which in many regards it is 

 closely akin. 



n. As to the very primitive characters of 

 Ceratodus which ally it to Mylostoma and 

 separate it widely from known Paleozoic 

 Lung-fishes. 



Eastman expresses his view as to the rela- 

 tionship of lung-fishes and arthrodires thus : 

 A primitive ceratodont (from which descend 

 directly Ceratodus and Neoceratodus) was the 

 progenitor of two side lines of fishes, one giv- 

 ing rise to more and more specialized lung- 

 fishes, the other to more and more specialized 

 arthrodires. Before the specialized line of 

 lung-fishes became extinct it gave rise suc- 

 cessively to such forms as Dipterus, Scau- 

 m,enacia, Phaneropleuron, Uronemus and 

 Ctenodu^: before the arthrodire line died out 

 it passed through phases represented in the 

 order Macropetalichthys, Romosteus, Mylo- 

 stoma^ Dinomylostoma, Ooccosteus, Dinich- 

 thys, Titanichthys. The fact that in all of 

 the mass of Paleozoic lung-fishes there is not 

 a suggestion of the hypothetical Ceratodus is 

 easily waived aside as due to the imperfection 

 of the geological record. And thus are re- 



jected DoUo's illuminating researches as to the 

 descent of the dipnoans. 



We may query, accordingly, the reasons 

 why the modern Ceratodus (Neoceratodus) is 

 assumed to be the primitive dipnoan — to say 

 nothing, for the present, of its kinship to the 

 arthrodira. And here Eastman's studies do 

 not appear adequate : Ceratodus, he points out, 

 has a cutting type of dental plates, it has a 

 diphy cereal tail (rather than heterocercal), 

 and it has fewer dermal head-plates. He 

 does not suggest, however, that we have at the 

 present time a fairly rich material of fossil 

 dipnoans, and he fails to indicate that in the 

 ceratodonts many characters conunon to the 

 early forms do not appear; in a word, East- 

 man does not explain clearly his paradox — 

 that we are to believe that these earliest 

 dipnoan characters should be regarded as 

 more modified than the structures of the 

 modern Neoceratodus. Indeed, the skeptical 

 reader remembers, on the contrary, that in 

 the earlier fishes the teeth are in the form of 

 tubercles, more or less shagreen-like in form 

 and arrangement; that in all the earliest 

 groups of true fishes, sharks, dipnoans, cros- 

 sopterygians, actinopterygians, there occur no 

 shear-like dental plates; that in the series of 

 definitely known lung-fishes beginning with 

 those in the Devonian, the tubercular teeth are 

 reduced gradually, and that only with the de- 

 velopment of their basal supports do there come 

 to be formed cutting dental plates. Moreover, 

 that this mode of evolution is the true one is 

 confirmed with singular clearness in the gen- 

 eral plan of the development of the teeth of 

 Neoceratodus itself — a great number of tuber- 

 cular denticles preceding the solidification of 

 their basal supports and the growth of bony 

 cutting ridges. In short, there is every reason 

 to conclude that the dental plates of Cera- 

 todus are derived from dental plates of 

 dipnoans of the paleozoic type, and there is 

 no tangible evidence that the dental plates of 

 the recent dipnoan picture the ancestral con- 

 dition. 



Again, who can doubt that the descent lines 

 of the dipnoans and the ganoids converge very 

 closely in the earlier paleozoic times? One 



