xviii. INDIAN TERTIARY AND POST-TERTIARY VERTEBRATA. 
latidens, as in the two figured in pL XXXI. figs. I, 4a of the “ F.A.S.” there is a 
tendency in the ridges to become somewhat higher, and to the development of 
distinct accessory tubercles ; and there is thus effected an almost complete transition 
to M. caiitleyi. On the other hand, it is but a step from the structure of the ‘ inter- 
mediate ’ molars of the latter to those of M, ijerimensis, although there is somewhat 
less probability of a coin^fiete transition existing between these two species. 
.The ‘ intermediate ’ molars of the present form very closely resemble those of M. 
lonf/irostris, but the liind talon and fourth ridge are longer and narrower, the cusps on 
the columns more numerous, and the accessory tubercle of the second inner column 
less developed: the third molar (judging by the one example) also differs by its 
long, straight, and narrow fourth ridge and talon. There can, however, be but little 
doubt that the two forms are allied. 
As the result of the foregoing comparisons it appears that the form here known 
as M. cautleiji (whether it is really entitled to rank as a species or merely as a variety) 
probably passed imperceptibly into M. latidens, and connects that species in one 
direction with M. lonyirostris, and in another Avith M. perimensis,^ which is itself allied 
to the M. sualensis and M. arvernensis group. It has been shown {infra, p. 143) to be 
not improbable that the tetralophodont M. perimensis (in which M. cautleyi^ was then 
included) is a more specialised form derived from the stock of the trilophodont M. 
pandionis^ and it appears that starting from the generalized ilf. angustidens {infra, p. 
18) two branches of more specialized forms may be traced, the one (in which there 
is a tendency of the valleys of the molars to be much blocked and their columns to 
be ranged alternately) passing through M. pandionis and M. perimensis to M. sivalensis 
and M. arvernensis ; and the other (in which the valleys remain more open and the 
columns of the ridges run straight across the crowns) passing through M. longirostris, 
M. cautleyi, and M. latidens to ElepJias clifti, and then through the higher stegodont 
forms to the true elephants. There are, however, signs that some of the middle forms 
of these two branches Avere mutually connected ; but the nature of this connection 
cannot be at present indicated, although it may be suggested that in regions like 
India where several species inhabited the same area there may possibly have been 
interbreeding between the different forms. The relationship apparently existing 
betAveen M. caiitleyi and ilf. longirostris may help to explain some of the difficulties 
alluded to on page 18 of this volume, and may indicate that the transition from the 
trilophodont to the tetralophodont type in this branch took place in the regions to 
the westward of India.^ 
As regards distribution, 31. cautleyi apparently occurs only in Perim Island and 
the western Punjab ; while M. latidens occurs not only in these regions and Sind, 
but extends through the typical Siwalik Hills, and thence to Burma and Borneo. 
1 In the first volume from not having distinguished II. cautleyi trom M. perimensis the writer regarded the latter as more 
nearly allied to M. longirostris than to M. arvernensis. 
2 In its short and wide m 3 M. cautleyi comes nearer to M. pandionis than does M. perimensis : the latter, however, agrees 
with the trilophodont species in its more abundant cement. 
3 Note 4 on page 1 8 should be confined to M. arvernensis. 
