FORT UNIO:S' OF CRAZY MOUNTAIN FIELD, MONT. 219 



sideration. The periptychid carpus is, in fact, "alternating", but 

 so is that primitive for and apparently fakly common among (other) 

 condylar ths. 



Osborn (1S98, pp. 177-179, 184-186) gives a long list of taligrade 

 characters as defining that group and common to pantolambdids and 

 peript5xhids. The great majority of these are, as he states, primitive 

 characters. As far as confirmed among the so-called Taligrada by 

 later research, they are also found to occur among or to be equally 

 typical of Condylarthra and hence have no bearing on the particular 

 question here considered. The only progressive taligrade character 

 given is ''molars triangular (tritubercular), selenodont", which is 

 decidedly untrue of the Periptychidae and opposed to Osborn's thesis. 

 Indeed, I cannot see that Osborn then advanced any actually valid 

 evidence in favor of the conclusion given, which has since become 

 taxonomic and phylogenetic dogma, largely on his authority. 



Matthew (1897) had already shown that even in the supposedly 

 typical condylarths, the phenacodonts, the early forms have alter- 

 nating, not serial, carpus and tarsus and that the Condylarthra there- 

 fore could not be defined and were not characterized by the mooted 

 primitive serial carpus and tarsus as had previously been supposed. 

 He therefore found no difficulty in retaining the Periptychidae in the 

 Condylarthra and gave a lucid and valid argument for doing so, even 

 though, as I now think, he minimized his evidence by much over- 

 stressing the resemblance of the periptychids to the pantolambdids 

 in limb structure and their difference from the phenacodonts and mio- 

 claenids in dental pattern. 



Upon transferring the Hyopsodontidae to the Condylarthra, Mat- 

 thew (1915b, p. 311) gave a long diagnosis of the Condylarthra, in- 

 volving the whole bodily structure. His intention at the time was to 

 exclude the Periptychidae, since he did so in earlier and later general 

 classifications, although this point was not then specifically mentioned, 

 since it was foreign to the fauna he was revising. It is therefore re- 

 markable and significant that his definition of the Condylarthra clearly 

 excludes the Pantolambdidae but applies exactly to, and hence includes, 

 the Periptychidae with a single exception: "tarsals serial." This one 

 point was, in fact, an error or lapsus, for the forms he explicitly meant 

 to include do not have strictly serial tarsals, and in some the approach 

 to the periptychids in this respect is very close. 



Matthew's subsequent defense of the collocation of Periptychidae 

 and Pantolambdidae was based almost entirely on the limbs, especially 

 on the astragalus. When evidence drawn from the dentition was at 

 variance with that drawn from the limbs or astragalus, he almost in- 

 variably followed the latter. Without quarreling with this principle 

 of research, it will appear below that the evidence is not necessarily 

 at variance in the present instance. The dental evidence certainly 



