106 Bulletin American Museum of Natural Hist(n-i/. [Vol. XXVII, 



The general aspects of this problem have been very fully dealt with bv 

 Cope (1896), Scott (1891), and Osborn (1902, 1904, 1905, 1910) and need 

 not be discussed here; but the universality of parallel evolution and the 

 confusing intermixture, in every form, of hereditary characters and homo- 

 plastic resemblances to distantly related forms, are conditions which seem 

 to call for some formulation and brief discussion of the principles which have 

 been applied in the following studies on mammalian phylogeny. 



Necessity for conducting pJu/loge^ietic research in accordance with the 

 strict rules of the inducti've process. — In the study of the genetic relations of 

 mammals there are very few maxims which are of universally deductive 

 application. Phylogeny is essentially an inductive subject, a reasoning by 

 analogy, which is the shifting sand whereon hypotheses and theories are 

 built. In general, the student must (1) concede nothing more than he is 

 forced to, (2) strive to separate probability from plausibility, (3) test his 

 hypotheses by the principle of negation, and (4) avoid explaining the little 

 known through the less known. Above all (5) he must strive to keep in 

 touch with all data bearing on the subject, (6) make constant reviews to see 

 that no pertinent fact has been omitted and (7) test again and again his basal 

 assumptions. 



These principles may indeed seem to be obvious councils of perfection; 

 but so much zoological study has been vitiated by the neglect of them that 

 it has come to be scarcely respectable to draw up a phylogenetic tree. 



Among the phylogenetic principles which have become fairly well estab- 

 lished the following seem to recjuire notice in the present study. 



Bearing of the imperfection of the record upon the interpretatioyi of mam- 

 malian phylogeny. — One of the cardinal postulates of the phylogenist should 

 be the well known imperfection of the geological record. There is much 

 evidence to show that many existing orders of mammals were already 

 represented in the Cretaceous or even earlier epochs, that is, that certain 

 of the more fundamental ordinal characters are older than the Tertiary; 

 and that therefore the points of separation of these orders occur where the 

 discovered record is extremely meagre. Many families also may have 

 acquired their family characters in some area as yet unopened by palseonto- 

 logical exploration, such as northern Asia. 



Palaeontology may be said to be little more than comparative anatomy 

 applied to faunas of different periods, ?'. e., to sections of the phylogenetic 

 tree taken at various planes. The true chronological succession of forms is 

 obscured by many factors, notably: (1) the imperfect knowledge of the time 

 equivalence of Tertiary mammal horizons in different continents (cf. Osborn, 

 1900), (2) the intermingling of immigrant and autochthonous elements in a 

 fauna through migration (cf. Deperet, 1908), (8) the intermingling of 

 persistent primitive and highly specialized forms in the same fauna. 



