176—22 INDIAN TERTIARY AND POST-TERTIARY VERTEBRATA. 
e.g.^ G. ornata^ C. venmta^ G. callirostris^ G. rivulata,^ and C. ventricosa.^ All these 
sjDecies agree with the Siwalik fossil in the absence of costal and (excej:*! in C. 
callirodris) vertebral keels. The last-named species conies nearest to the fossil in the 
form of tlie first three vertebral plates, but the second plate is not so much contracted 
posteriorly : in the contour of the shell C. ventricosa makes a nearer approach to the 
specimen under consideration, but all the American species are more depressed. 
As there seems no doubt of the sjjecific distinctness of the present form from 
any Old World species of Clemmys^ and as it is apparently equally distinct from, 
although more nearly allied to, those of N. America, it may rank as a new species, 
to which the name of C. pimjabiensis may be assigned. 
The total distinctness of C. pimjabiensis from the existing Indian emydines, and its 
a2325arent kinship to those of N. America, is a remarkable and unexpected fact in the 
distribution of this group of tortoises. 
B. Carapace with three continuous keels. 
. Clemmys, sp. 5. 
(Allied to Glemmys trijuga [Schweigg.]) 
Bhell. — The evidence of the existence of this fifth species of Siwalik Glemmys is 
afforded by a shell from the Siwalik Hills preserved in the Cautley collection of the 
British Museum, and represented of one-half the natural size in plate XXI. figs. 
4. 4a. 4b. It comprises the nearly entire caraj^ace, and the anterior j)art of the 
jDlastron ; the latter having been crushed in, but exhibiting the gular and postgular 
plates (fig. 4b), and a portion of the pectorals ; on the right side a jjart of the suture 
between the abdominal and 23ectoral plates, and that between the former and the 
marginals is also jDreserved. The carajDace has lost most of the nuchal and part of the 
1st marginal plates, as well as the greater portion of the joygal and hinder marginals ; 
but sufiicient remains to show that the j)ygals were not united. The plastron is 
united by bone to the carapace, showing that the sjjecimen cannot belong to 
Terrapenej Cyclemys^ or Pyxidea : the difference in the form of the gular 2)Htes 
distinguishes it from a fenqale of Geoemyda, and as it is evidently not a Pangshura 
and almost certainly not a Datagur, it may be pretty safely referred to the present 
genus. 
The gular j^lates (fig. 4b) are of rather small size, the length of the intergular 
suture being only about one-and-a-half times that of the one between the jDOstgulars. 
The carapace is subovate, somewhat depressed, slightly flattened superiorly as far 
as the 5th vertebral plate (which is obliquely placed) and bears three indistinct, low^ 
and api^roximated keels, which are continuous throughout the length of the shell. 
In profile (fig. 4a) the anterior jDortion presents a gentle upward inclination from the 
margin to the vertex, while posteriorly there is a sudden ascent from the margin. 
1 FjflJe Gray, “ Catalogue of Shield-lieptiles,” pt. I. pi. XII. 2 Ibid, pi. XIIa. 3 Ibid, pi. XIIb. 
■i Ibid, pi. XI. 5 Ibid, pi. XIV. 3 Including- Unoni, Gray. 
/ 
