FORT UNION OF CRAZY MOUNTAIN FIELD, MONT. 231 



tidae and Mioclaenidae, but he saw that Haplomylus could not be 

 ancestral to Hyopsodus and was not clearly derivative from any 

 known Paleocene form, so that the evidence was inconclusive. 



The discovery, since Matthew, of several Middle and Upper Paleo- 

 cene genera that are clearly related to Mioclaenus and Ellipsodon but 

 that approach Hyopsodus much more closely than do those two 

 genera has much altered the conception of this family. It is now seen 

 that in the Paleocene there are a less Hyopsodus-\ike and a more 

 Hyopsodus-like group. The distinction of the Hyopsodontinae, with 

 Hyojjsodus only, depended in part on progressive characters, sure to 

 become uncharacteristic when forms of intermediate age and structure 

 were found, and in part on what appear to be valid phyletic characters 

 which separate Hyopsodus from one group of Paleocene genera, but 

 associate it with the other. On present data, it seems preferable to 

 base supergeneric classification on these latter characters and to draw 

 the line not between the Paleocene forms and Hyopsodus but between 

 those of the Paleocene forms that are less and more like Hyopsodus, 

 grouping that genus with its closer relatives among the older genera. 



Subfamihes drawn upon this basis are defined below. This arrange- 

 ment is still only tentative, and it is clear from the discussion of 

 generic relationships above that a great deal must yet be learned 

 before a really well-founded classification witliin the family will be 

 possible, but the new arrangement perhaps represents a step toward 

 this end. The most doubtful points, as regards the forms now known, 

 are the affinities of the more atypical species placed in Ellipsodon and 

 of Litaletes, the true place in the system of the rather isolated genus 

 Protoselene, that of the apparently aberrant Haplomylus, and the 

 relationships of Phenacodaptes. 



The new evidence substantiates without greatly altermg the grounds 

 for considering the hyopsodontids as condylarths, sufficiently set 

 forth by Matthew. Discovery of intermediate forms makes the 

 family more coherent than it seemed to him and improves the evidence 

 for considering the relatively well known Hyopsodus as indicative of 

 the affinities of the Paleocene genera, and so strengthens his conclu- 

 sions. The resemblance of the early hyopsodontids to the dichobunids 

 in the dentition is so remarkably close that it is difficult to ascribe it 

 entirely to convergence. Although the known skeletal parts are not 

 of artiodactyl type, it is quite possible that some branch of the earliest 

 hyopsodontids did give rise to the Artiodactyla, but this can be proved 

 or disproved only by further discovery. Even if this should prove 

 to be the case, the hyopsodontids as a whole would probably be best 

 classified as Condylarthra, since they had the general characters of 

 that order, and retained them after the ancestral artiodactyls were 

 distinctly differentiated in the skeleton. 



119212—37 16 



