Flora of Narcondam and Barren Island. 
61 
and the various hot springs in the valley of Assam, like those in the 
ISTamha Forest,* are examples of this series, which forms a continuous 
line parallel on its western side to the tertiary ridge referred to, just as 
the true volcanoes, to the line of which Barren Island, Narcondam and 
Popah belong, are parallel to it on the east.f 
Whether they belong to that particular group of volcanoes known 
as the Sunda Range, or not, there is no doubt that Rarcondam and 
Barren Island belong to the general volcanic system extending from 
the Kuriles, through Japan and the Philippines, to Malaya—a system 
of which the Sunda Range itself forms but a portion. Like the other 
members of this system, these peaks are situated, not on, but just within, 
the margin of the continental elevation forming Eastern and South- 
Eastern Asia, wherever this rises abruptly from great ocean-depths; the 
main difference between them and most of the peaks of the system is 
that, whereas the space between the edge of the continental area and 
the line of volcanic activity is in other cases sub-aerial, that space is here 
for the most part sub-marine. This space forms, in the case of Sumatra, 
the main body of the island—the volcanic line being much nearer the 
eastern margin—and the rocks of which it is composed include all those 
that go to form the islands of the Nicobar Group; these rocks appear 
once more, not in the main chain of the Andamans, but in the small 
islands to the east of South Andaman (north-east of Port Blair) known 
as “ The Archipelago.’’^: Neither in, nor opposite, the Nicobars is there 
any trace of the complementary volcanic ridge ; to the east of this “ Archi¬ 
pelago,” however, it is indicated by Flat Rock and Barren Island. 
Not only is the volcanic line of Sumatra absent from the Nicobars, 
but no trace has yet been found in that group of the sandstones of the 
Arracan hills, which are prolonged into the main chain of the Andamans 
and which reappear in the Nias, The result, therefore, is that the 
Arracan-Sumatra chain, in place of constituting a single ridge consists 
* Prain : Proceedings As. Soc. Bengal, 1887, p. 201. 
t The reasons for thinking that the northward prolongation of the Sunda 
Range has not crossed the Arracan-Andaman ridge are, therefore :— 
1. That the volcanoes on the west side of that ridge, which are supposed to 
continue the Sunda line, are of a different type from the volcanoes of the Sunda 
Range. 
2. That these western volcanoes in Ramri belong to a system of vents of the 
same type as themselves, characterised by a linear distribution parallel to the 
western base of the Arracan-Andaman tertiary ridge. 
3. That the Sunda Range is continued northward by a series of vents of the 
same type throughout, the character of linear distribution parallel to the eastern 
base of the Arracan-Andaman tertiary ridge being maintained unaltered. 
1 Oldham : Records of the Geol. Survey of India, xviii., 141. 
275 
