492 THE STRUCTURE AND RELATIONSHIPS OP ANCODUS. 
1„ AimJm Hie Incisors arc spaced well apart, and have large, spatulate crowns; 
the calls are ordinarily smaH, except in J. Up<.kynchus, and the lower one 
dillers hat little from the incisors in shape. The premolars are much snnpler than 
those of Ore, dm,, not havins the ridges and tubercles developed on the internal 
faces The upiwr molars are very different from those of the last-named genus ; 
not only have they retained the pmtoconule, but the, have very broad, low crowns 
with deeply concave outer crescents, which project far toward the median part of 
the tooth, and with very prominent and tuberous external buttresses or styles. On 
the lower molars the crescents are very much higher, somewhat thicker, more 
conical and less completely crescentic in shape. 
Agriochcerus combines some of the features of the other two genera with 
characteristic peculiarities of its own. It agrees with Oreodon in the character of 
the canines and caniniform first lower premolar, and in having lost the protoconule, 
thus making the upper molars tetraselenodont. Aside from this, however, the 
molar pattern is very much more like that of Ancodtis, though the cusps of the 
lower ones are less elevated. The premolars are, for the most part, less compli- 
cated than in Oreodon. Peculiar features are the more or less complete reduction 
of the upper incisors, and the molariform pattern of the last premolar in each jaw. 
Protoreodon nearly represents the common term from which all three types of 
dentition may easily have been derived, and though it has already assumed too 
many oreodout features to be actually the real starting point from which the later 
genera diverged, it greatly reduces the gaps between them. The incisors and 
canines are like those of Oreodon,, the premolars have the simple compressed coni- 
cal form which recurs in Ancodus and Agriochcerus ; the upper molars still retain 
a well-marked protoconule, like that of Ancodus, and the lower molars are almost 
exactly like those of Agriochosrus. The particular interest attaching to Protore- 
odon consists in the proof which it gives that the oreodonts were derived from an- 
cestors with quinque-tuberculate superior molars, having the fifth lobe in the ante- 
rior half of the crown, and this brings them into relationship with the anoplothe- 
rioids, anthracotherioids, etc., as distinguished from the dichobunids and csenothe- 
rioids, to which it seems probable that, as Sclilosser has suggested, the existing 
lines of ruminants should be traced back. 
1 he character of the vertebral column is very similar in both Ancodus and 
Oreodon, es])eciii\\y if the species of the former which occur in the Protoceras-beds 
be used in tlie comparison, the differences being merely such matters of detail as 
are incidental to the great difference of stature. In the later species of Ancodus 
(e. g. A. brachyrhynchus) the odontoid process of the axis is, like that of Oreodon 
and Agriochcerus, neither conical nor spout-shaped, but half-way between the two 
patterns. Ancodus americanus from a lower horizon, however, has a fully conical 
odontoid, and this shows that the resemblance of A. brachyrhynchus to Oreodon in 
this respect is a case of parallelism, and has been acquired within the limits of the 
genus. 
The scapula, which as yet is very imperfectly known in Agriochcerus, is alike 
