) 
B82 NOTES ON THE CANID® OF THE WHITE RIVER OLIGOCENE. 
invariably became separated from the skull on fossilization. At the bottom of the fossa 
(7. e., When the skull is turned with its ventral surface upward) is seen the exposed 
periotic, or petrosal, which is only partially overlapped and concealed by the tympanic. 
Such an arrangement is far more primitive than that which is found in any other known 
member of the canine series, and is not easy to interpret. A clue to its meaning may, 
however, be found in the mode of development of the bulla in the recent Canidw. Here, 
as is well-known, the structure consists of an anterior membranous and posterior carti- 
laginous portion, which eventually ossify and coalesce into a single bulla. Reasoning 
from this analogy, we may infer that in Daphwnus the bulla was also composed of two 
portions, but that only the anterior chamber was ossified, the posterior one remaining 
cartilaginous. Communication between the two chambers was provided for by the space 
which separates the hinder edge of the anterior chamber from the petrosal. If this 
interpretation be correct, it supplies an interesting confirmation of the results derived 
from the ontogenetic study of the recent genera. At all eyents, it seems much more 
probable that we have to do here with a primitive rather than a degenerate structure. 
The parietals are large and roof in most of the cerebral fossa; they are much less 
convex and strongly arched than in Canis, in correspondence with the smaller size of the 
cerebral hemispheres, and posteriorly the depressions behind the hemispheres are much 
larger and deeper. As already remarked, the sagittal crest varies in the different species, 
and is much thicker and more prominent in D. vetus than in D. hartshornianus. The 
frontals are more or less damaged in all the specimens and in none of those at my disposal 
is it possible to determine the posterior limits of these bones, though from the position of 
the postorbital constriction. we may confidently infer that they formed a smaller proportion 
of the cranial roof than in the modern members of the family. The supraciliary ridges are 
feebly developed, especially in D. hartshornianus, and the postorbital processes are like- 
wise much less prominent than in most of the recent dogs; from this process a ridge de- 
scends downward and backward to the optic foramen, which, though not prominent, is yet 
more so than in Canis. The frontal sinuses are large and yet in spite of them the forehead 
is nearly flat, both longitudinally and transyersely, with a very shallow depression along 
the median line. The nasal processes of the frontals are long, narrow and pointed, 
and are separated by only a short interval from the ascending rami of the premaxillaries. 
The squamosal is of moderate size and differs only in subordinate details from that 
of Canis. One such difference is the presence of a broad shelf-like projection, the pos- 
terior extension of the root of the zygomatic process, which overhangs the auditory 
meatus and is doubtless to be correlated with the lesser breadth and conyexity of the 
brain. The glenoid cavity is like that of the recent species, but has a much more 
distinct internal boundary, due to an elevation of the squamosal at that point. ‘The 
