THE MAMMALIA OF THE DEEP K1VEK BEDS. 115 



of Ancliiilurium had entered upon a course of development which was not in the 

 direction of the typical horses, but leading away from them, and that in consequence 

 the genus had no place in the direct line. 



(2) If we may assume that the coalescence of the meso- and ectocuneiforms in 

 the tarsus is really characteristic of the genus, we shall have a further reason for 

 denying AncJiitherium a place in the direct ancestry of the horses, for it seems 

 unlikely that the modern condition should have been already attained in MesoJiippus, 

 lost in Ancliiilierium, and reacquired by the subsequent genera. But, in view of the 

 uncertainty as to the typical character of this structure, we cannot insist strongly 

 upon it. 



(3) The very curiously elongated and flattened hoofs of this genus also militate 

 against the view that it belongs in the direct line, since in the change of MioJiippm to 

 Desmatippus they do not represent one of the stages of the transition which v/e 

 should expect to find. 



(4) The very large size of both the known species of AncJiitJierium, one European 

 and the other American, would seem to indicate that this is characteristic of the 

 genus. This size much exceeds that of the forms which, on the hypothesis that 

 AncJiitherium belongs in the main series, must be regarded as its successors, and 

 such alternations in bulk are unlikely. 



(5) There is no vacancy in the direct equine phylum which AncJiitherium can 

 fill, as the change from Miohippus to this genus, though of a different kind, is hardly 

 less in amount than that from Miohippus to Desmatippus, and to insert AncJiitJierium 

 in the series would be to assume a view of zigzag development, which, as to amount, 

 is unnecessary and unwarranted. As we shall see later, a certain degree of such 

 alternating advance and retrogression very probably does take place, but not to such 

 an extent as this hypothesis would involve. It might be thought that the occurrence 

 of AncJiitherium in the same horizon with the more modernized genera, Protohippus 

 and Desmatippus, would be a further argument for excluding the first-named genus 

 from the phylogeny. This fact must, of course, be allowed some weight; but as it 

 is, perhaps, a case of the survival of an older form, just as Desmatippus very prob- 

 ably is, no great importance can be attached to it. Such cases must usually be 

 decided upon morphological grounds. 



If the view as to the systematic position of AncJiitJierium here contended for be 

 correct, it follows that those features in which this genus approximates the modern 

 forms more closely than does MioJiippus are phenomena of parallelism. As such, 

 these structures deserve careful attention. Assuming the possibility of parallel 

 development, we might on a priori grounds lay down the general principle that the 



