EASTMAN: STRUCTURE AND RELATIONS OF MYLOSTOMA. 5 



to Ctenodipterines on the other, although conforming in certain respects 

 more closely than either to the hypothetical common ancestor from 

 which all three types — Ceratodonts, Arthrodires, and Ctenodipterines 

 — have been derived. 



The position maintained in this last communication is adhered to, 

 and it is believed that sufficient evidence has now been accumulated 

 to sustain its correctness. Heretofore, in default of positive evidence, 

 writers have been unable to demonstrate the truth of any one of the 

 various conjectures put forward to explain the nature of Arthrodires. 

 However plausible one or another of these may have appeared, however 

 firmly they have been insisted upon, it must be remembered that a 

 suggestion remains only a suggestion, and an hypothesis an hypothesis, 

 until its correctness is clearly proved. Not without reason is it observed 

 in one of the Socratic dialogues, that " mere beliefs and opinions are, 

 like the statues of Daedalus, runaway things ; not until they have been 

 tied down by the chain of causal sequence do they stand fast and 

 become in the true sense knowledge." (Meno, 159 D). 



What constitutes "reasoned interconnection," as Plato calls it, in the 

 present case, lies in the recognition of actual, definite, and precise homol- 

 ogies between Arthrodires and typical Dipnoans, which have hitherto 

 escaped attention. That the significance of certain Arthrodiran char- 

 acters has not been fully appreciated heretofore is due in large measure 

 to the lack of sufficiently instructive material ; and in part, also, to 

 wrong interpretation of existing materials. By a fortunate chance the 

 former of these deficiencies is now remedied, valuable enlightenment 

 being afforded by the type of a new genus of Arthrodires from the 

 Portage of western New York, presently to be described under the 

 name of Dinomylostoma. It is hoped, also, that the second of these 

 difficulties may be removed by means of a novel interpretation of the 

 jaw-parts of Coccosteans and Mylostomids, such as is hereinafter set 

 forth. Altogether, it would appear that a sound basis is now provided 

 for upholding the following general propositions : — 



1. Cranial roof- plates have undergone corresponding reduction and 

 have become arranged after essentially the same pattern, both in 

 Arthrodires and primitive Ceratodonts. 



2. Neoceratodus recalls throughout its entire organization, save only 

 for the absence of body armoring, the principal features of Arthrodires ; 

 resemblances which form too large an aggregate to be explained through 

 parallelism. 



3. It is. impossible to regard Neoceratodus as the degenerate de- 



