Bulletin American Museum of Natural History. [Vol. XIV, 



the present writer that group and race would more truly ex- 

 press their intent. By the former method the unspecialized an- 

 cestral types of several families are placed in a separate family, 

 from which the later families are derived. This is on the ground 

 that the distinctions between the different lines of descent 

 were not at first wide enough to be of family value, and that the 

 different primitive types resembled each other more than they 

 did any of the later specialized types. The other method of 

 classification divides up these primitive types among the various 

 specialized families to which they are ancestral or approximately 

 ancestral. The group conception — the Latin familia = house- 

 hold — is, of course, the original use of the term ; the race con- 

 ception — the English /rtw//j/ — has been introduced to meet the 

 new conditions brought in by the doctrine of evolution and the 

 development of palaeontology. 



It seems to the writer that either principle, used exclusively, re- 

 sults in obscuring, or at least imperfectly indicating, the real 

 relationships the expression of which should be its chief pur- 

 pose. The group method ignores parallelism, and fails to properly 

 emphasize the lines of descent. The race method equally fails 

 to emphasize the near relationship of the primitive root types, 

 and in practice causes much confusion and apparent variance of 

 opinion by the attempt to divide into different families species 

 among which the distinctions have not yet become of generic 

 value. 



Among the fossil Mammalia these two methods are used in 

 very varying proportion. In some groups, such as the Perissodac- 

 tyls, the divisions have been drawn ' vertically,' all the ancestors 

 of a family being placed in that family, so that the primitive Peris- 

 sodactyls of the Lower Eocene, exhibiting no more difference in 

 the sum of their characters than the different species of mod- 

 ern Rhinoceros, are divided among six different families, the 

 typical forms of which are extremely different from any of these 

 primitive forms ; and the various species are moved about from 

 one to another of these families with the utmost facility by every, 

 author who attacks the problem anew. In fact, constant specific 

 distinctions are not always easy to find among them. In other 

 groups, such as the Creodonta, all the ancestors of the modern 

 types, together with those ancient types which have left no 



