July 12, 1907] 



SCIENCE 



49 



may even be doubtful wbether certain of these 

 genera were ganoids rather than lung-fishes, 

 and close examination of the known struc- 

 tures of these forms has led every observer, as 

 far as I am aware, to postulate the closest 

 kinship between the two groups. From these 

 early types, upward, one may trace in the 

 fossil lung-fishes the dermal plates of the 

 head-roof becoming less numerous, lighter in 

 texture, and deeper in position, losing com- 

 pletely their primitive tubercle-studded sur- 

 face. From Ceratodus (as Teller's figures in- 

 dicate) to Neoceratodus there is a marked 

 step in this direction, and from such a condi- 

 tion only can one understand the curiously 

 reduced dermal head-roof of Protopterus and 

 Lepidosiren. Why, accordingly, should we 

 believe, in the face of this kind of evidence, 

 that the condition of the head-roof of Cera- 

 todus is more primitive than that of the early 

 ganoids and dipnoans conjoined? There is 

 certainly adduced no concrete evidence for 

 such a reactionary view. Eastman's final 

 evidence as to the ancestral nature of 

 Neoceratodus, as far as I am able to find, is in 

 the shape of its caudal fin: it is diphycercal 

 rather than heterocercal. Dollo has shown, 

 on the other hand, that the earliest dipnoans 

 (ganoids and sharks as well) are heterocercal, 

 and that it was only through the paleonto- 

 logical series, which he carefully depicts, 

 that diphycercy was attained in the modern 

 lung-fishes, as an eel-like adaptation to living 

 in a ' muddy bottom — an evolution in the 

 process of which the dorsal and anal fins 

 became merged with the caudal. This con? 

 elusion of Dollo is based upon such strong 

 testimony that it can hardly be disproved 

 merely by the assumption that a priori 

 a diphycercal caudal fin is more primitive 

 than a heterocercal one! In short, we can 

 find in Eastman's studies no ground for 

 making the stock of Neoceratodus an ancestral 

 one; there is, indeed, no reason evident why it 

 should not have descended from an ancestor 

 resembling TJronemus or Phaneropleuron. 



III. Mylostoma as a Primitive Arthrodire, re- 

 lated to a Oeratodont Lung-fish. 

 Mylostoma differed little from its contem- 



porary arthrodires. In its gnathal plates, 

 however, it had evolved restricted crushing 

 surfaces instead of the long tubercle-studded 

 jaw-rims of Diplognathus, Trachosteus, Selen- 

 osteus or Coccosteus. Diniclithys, indeed, 

 shows transitional characters, for the tubercles 

 of the anterior reaches of the jaws are ground 

 away when the jaws attain a shear-like ac- 

 tion, and the gnathals of Dinomylostoma - 

 show a still nearer approach to the pavement- 

 like surfaces of Mylostoma. In short, there 

 is evidence that the arthrodira during their 

 extraordinary evolution gave rise to a series 

 of forms whose dental characters ranged from 

 tuberculate to pavement-like — a line of evolu- 

 tion which, it will be recalled, is paralleled in 

 other groups of fishes — sharks, ganoids, 

 teleosts, and, as above noted, dipnoi. Now 

 since the time of the classical studies of O. 

 Hertwig (1876) on the origin of the bony 

 plates of fishes, there has been found no good 

 reason to doubt that the tuberculate condition 

 was the ancestral one, and it follows, there- 

 fore, that until strong reasons to the contrary 

 be adduced, we can safely assume that the 

 same law of development holds true in the 

 case of the arthrodira. That is to say, that 

 the crushing plates of Mylostoma are second- 

 ary, not primitive. Eastman, however, con- 

 tends that since Mylostoma resembles Cera- 

 todus, it is therefore primitive. But if, 

 as we have indicated above, there is little 

 reason to regard Ceratodus as primitive, it 

 is clear that the affinities of Mylostoma 

 must be determined by comparison with 

 kindred arthrodira. It might be pointed out, 

 finally, that the great majority (possibly 

 eight out of ten) of the genera of which 

 jaw plates are known, bear tuberculated 

 dental plates, including the earliest knovsm 

 arthrodires. , And this is naturally inter- 

 preted in favor of the modified nature of 

 Mylostoma, for thus historical evidence sup- 

 ports the findings of comparative anatomy. 



If, now, the foregoing objections to East- 

 man's conclusions are valid, it is quite clear 

 that the general question of the affinities of 

 the arthrodira is just as doubtful as ever. 

 Eastman, emphasizing the dipnoan characters- 



