98 AMERICAN MESOZOIC MAMMALIA 



represented fauna is in fact very inadequately known. The specimens consist for the 

 most part of isolated teeth which cannot be associated into natural genera in the 

 majority of cases. A few fragments of jaws are known, but even these are inadequate 

 for the solution of many of the more detailed problems of taxonomy. The attempted 

 revision of the upper Cretaceous faunas is thus on quite a different basis from that of 

 the Jurassic fauna. With the latter, whatever its imperfections, one is at least dealing 

 for the most part with jaws with their included teeth and direct comparisons between 

 the established genera are possible in most cases, but in dealing with the Cretaceous 

 mammals even the characters of two consecutive teeth of a single genus cannot be 

 determined in many instances. 



The history of our knowledge of the faunas has also added difficulties. This 

 knowledge is chiefly due to O. C. Marsh and H. F. Osborn. These two workers ap- 

 proached the problem from diametrically opposed viewpoints. As the fragmentary 

 Lance collections came to him, Professor Marsh proceeded to name and publish them 

 on an analytical basis, that is, he applied names not necessarily to distinct animals but 

 to different t)^es of teeth. Under the circumstances there was much to be said for this 

 procedure. Admitting its artificial nature, yet it would have given an adequate nomen- 

 clature which would have served as an admirable basis for later revision if it could 

 have been carried out systematically. In practice, the method broke down, indeed its 

 apparent basis of analysis was perhaps not formally recognized. The number of differ- 

 ent tooth types became so great, that Marsh felt forced to attempt synthesis in some 

 cases, that is, to refer different types of teeth to one genus or to one species on the 

 grounds that they possibly were different parts of the same type of dentition. Marsh 

 himself thus recognized the impracticability in this case of what we must perceive to 

 have been his fundamental approach, but the inadequate data at hand for synthesis 

 made correct association in the limited genera which he had established impossible 

 and his attempts at such association thus led rather to further confusion than to clarity, 

 although this was due to nature of the material. 



Marsh's work was done rapidly, as the material came in, lot by lot, from the field 

 with the expressed hope that rapid publication and illustration would compensate for 

 the errors which Marsh himself knew to be inevitable under these circumstances. Pro- 

 fessor Osborn, with Marsh's work before him, attempted a synthetic revision. His aim 

 was to abandon any attempt to give a distinct name to each distinct type of tooth and 

 instead to give names only to whole animals, that is to reconstitute the dentitions scat- 

 tered parts of which had been given different names by Marsh in many cases. This 

 method, while simpler and giving more enduring results, also had its shortcomings. 

 In order to make it workable at all, the modern conception of genus and of species 

 could not be applied. Thus it is evident that the material referred by Osborn to Ptilo- 

 dus includes many species which would be seen to be quite distinct were good jaws or 

 complete dentitions of each available, or even if the whole taxonomy were based on the 

 variations in homologous teeth, P4 or M^ for example. But in attempting to base a 

 taxonomy on all the material, including isolated teeth from all parts of the dentition, 



