118 INSECTIVORA. 



I liave never seen the skeleton of Ptilocercus, which Mivart refers to the family 

 Tupaiidcs along with Eylomys ; but as an inspection of the skeleton of the latter has 

 not verified his presumption that it agreed with Tupcda in having its tibia 

 and fibula distinct, it is impossible to say what characters Ptilocereus may present 

 in that respect. 



The vertebral column of Tupcda has an upward curve in the cervical portion, 

 and the dorso-lumbar region is curved convexly — an arrangement of the vertebrae 

 which would lead to the conclusion that the animal was in the habit of sitting 

 with its body raised as the squu-rels do, and which I have observed to be the 

 case. The leading features which first present themselves are : the moderately 

 broad, thin, spinous plate of the axis, which appears large owing to its close 

 apposition to the narrow and rather short spinous processes of the cervical and 

 dorsal vertebrae; the depressed, overlapping spinous process of the three dorsal 

 vertebrae bearing free ribs, and of the Imnbar with their well-developed trans- 

 verse processes; the rapid enlargement of the vertebrae in the lumbar region; 

 the rather small sacrum, consisting of three vertebrae; and, lastly, the long and 

 tapering tail. 



There are 13 dorsal and 6 lumbar vertebrae with 23 caudal vertebrae or more. 

 The spinous process, if we except the axis, is fully developed in the cervical 

 region. It first becomes distinctly visible in the 5th vertebra, increasing in size 

 to the 7th. The spinous process of the 1st dorsal is nearly erect, whilst the 

 2nd is slightly stronger and broader antero-posteriorly, but about the same length 

 and directed shghtly backwards. The 3rd is still stronger and a little more 

 lengthened and backward in its direction, while the 4th is almost its fellow. The 

 5th is not so broad from before backwards as the 4th, which it exceeds in length ; 

 and the 6th shows a distinctly more marked backward inclination and is also 

 narrower than the process on either side of it. The 7th, 8th, and 9th processes 

 increase gradually in antero-posterior expansion, but diminish in length, the 9th 

 being almost triangular, with a broad base. The lOth process is smaller than the 

 9th, with a sharp triangular point with its apex directed forwards. The 11th is 

 quite distinct from those preceding it, and has its apex extended from before back- 

 wards, and directed forwards over the lower two-thirds of the hinder margin of the 

 10th process. It is short and rather broad antero-j)osteriorly. The 12th and 13tli 

 spinous processes are low and broad from before backwards, and overlap the vertebrae 

 in front of them. The first lumbar process resembles the last mentioned, but 

 the one siicceeding it is not so depressed, and is every way larger. As the processes 

 are traced backwards to the last lumbar vertebrae they become larger and more erect, 

 although still directed fonvards. The spinous process of the 1st sacral is low 

 and ridge-Uke, but in the 2nd vertebra it is moderately large and bent slightly 

 forwards. In the 3rd sacral it is lower than in the last and slopes faintly backwards, 

 and the process, although moderately developed in the 1st caudal, rapidly diminishes 

 in size and disappears in the 3rd, being continued to the 6th as an obscure ridge 

 wliich is lost in the 7th segment. 



