390 !'liOCEEUI\G>! or the XATIOXAL museum. vol. 4!). 



The rear of the maxilla on each side forms a great capsule, in which 

 is enclosed the developing hindermost molar. This protrudes 

 upwardly into the front of the temporal fossa and almost into the 

 rear of the orbit. It has a length of nearly 80 mm., a height of 53 

 mm., and a- width of 48 mm. 



The premaxillae are seen on the upper surface of the skull as a 

 pair of lance-like processes, one at each side of the nasal opening, 

 passing backward to become wedged in between the nasal and the 

 ascending process of the maxilla. At the rear of the nasal opening 

 each had a width of 30 mm. For our laiowledge of the front part 

 of these bones we must depend on the Japanese authors who have 

 been already referred to. In their specimen the premaxillae formed 

 the tip of the snout and extended back in each side of the nasal 

 opening. From the front of the snout to the front of the nasal opening 

 was a distance of 70 mm., from which fact we may conclude that the 

 distance was about 35 mm. in Desmostylus liesferus. The amount 

 missing in our specimen must be close to 75 mm. From the front 

 to the rear of each premaxilla was then about 145 mm. 



It is evident from Yoshiwara and Iwasaki's figures that the pre- 

 maxillae, while retaining their width, thinned out in front to a 

 transverse edge. They could hardly have been armed in front 

 with teeth of any considerable size. 



In the Oregon specimen the premaxillo-maxUlary suture appears 

 to be preserved on the palatal surface. Near the midline it turns 

 back a distance of 20 ram. and ends at the midline. As the bone 

 is injured at the midline, nothing can be determined regarding the 

 anterior palatine foramina. 



As already stated, Marsh based the genus Demostylus on teeth. 

 He described them as being composed of a number of vertical columns 

 closely pressed together, and in adult animals firmly united at their 

 bases. He stated that in immature teeth the columns are nearly 

 round and loosely united, but as they increase in size they press 

 together and become more or less polygonal in cross section. These 

 statements appear to be wholly correct. He says further that 

 before being worn they have their summits smooth and convex, but 

 after some use the center of each column presents a rounded eleva- 

 tion, such as is shown in his figure. This appears to be an error. 

 The hindermost tooth of the Oregon specimen had oidy just come 

 through the bone and had certainly not come into use ; but the sum- 

 mits of the columns have exactly the structure described by Marsh 

 (pi. 57, 23). The three columns which he figured are certainly those 

 of an unworn tooth. 



As to the number of columns in each tooth Marsh was uncertain; 

 but he thought that there were indications of at least twelve or 

 fifteen. Lrom what is at present known tiiis conclusion is erroneous. 



