MODES OF MAMMALIAN EVOLUTION 649 



of the science, has abeady taught us much concerning the laws 

 of inheritance and the manner in which new characters arise, 

 and no one can venture to fix the limits of its possible results. 

 On the other hand, it does not seem likely that the larger 

 problems of relationship and classification can be solved by 

 this method, because of the brief time which the shortness of 

 human life allows for the experiments. 



(4) Palaeontology suffers from the drawback that much of 

 the past history of life is irretrievably lost, and even when the 

 record is remarkably complete, as it is for certain chapters 

 of the history, the material is but partially preserved. With 

 such rare exceptions as are of Uttle practical importance, only 

 the hard parts, bones, teeth, etc., are retained and the soft 

 parts completely destroyed. Nevertheless, Palaeontology has 

 the preeminent advantage of offering to the student the actual 

 stages of development, and thus, to recur to the simile of lan- 

 guage, has preserved original documents and in the true order 

 of succession. It is true that it is weU-nigh impossible to re- 

 construct a phylogenetic series of ancestor and descendant, 

 imaffected by theoretical preconceptions, and the differences 

 which arise in the interpretation of undisputed facts are caused 

 by divergent beliefs concerning the actual course of the evolu- 

 tionary process. If final and definitive results are ever to 

 be reached, it must be through the cooperation of all the methods 

 of research, and such results must be able to stand the tests 

 appUed by every soimd method. On the other hand, the study 

 of those phylogenetic series which are generally accepted as 

 well established, should furnish lis with some fairly definite 

 information as to the modes in which development has operated 

 in the past, since the order of succession in time fixes a limit 

 to the rearrangement of related series. Some of the conclusions 

 thus suggested may be stated here. 



I. One of the most fxmdamental problems concerning the 

 course of development is that which deals with parallel and 

 convergent evolution. The term parallelism impUes that forms 



