408 CORALS AND CORAL ISLANDS. 
extinct Mammalia are found which must have inhabited a far 
more extensive area subsequently broken up by subsidence.” 
The position of the lonely Bermuda atoll confirms these 
deductions. Its solitary state is reason for suspecting that 
great changes have taken place about it; for it is not natu- 
ral for islands to be alone. The tongue of warm water due 
to the Gulf Stream, in which the Bermudas lie, is narrow, 
and an island a hundred miles or more distant to the north- 
east-by-east, or in the line of its trend (p. 219), if experiencing 
the same subsidence that made the Bermuda land an atoll, 
would have disappeared without a coral monument to bear 
record to its former existence. Twenty miles to the south- 
west-by-west from the Bermudas, there are two submerged 
banks, ten and twenty-four fathoms under water, showing 
that the Bermudas are not completely alone, and demonstrat- 
ing that they cover a summit in a range of heights; and it 
may have been a long range. 
In the Indian ocean, again, there is evidence that the 
coral-island subsidence was one that affected the oceanic area 
more than the adjoining borders of the continent, and most, 
the central parts of the ocean. For, in the first place, the 
Archipelago of the Maldives narrows and deepens to the 
southward (p. 186). Further, the large Chagos Group, lying 
to the south of the Maldives, contains but very little dry 
land in any of its extensive reefs, while some of them, includ- 
ing the Great Chagos Bank, are sunken atolls. Again, still 
other large reefs nearly bare lie to the southwest of the 
Chagos Bank, and submerged banks exist in the seas north- 
east and east of northern Madagascar. 
The probability is, therefore, that both the central Atlan- 
tic and Indian Oceans included regions of subsidence like the 
central Pacific, and that the absence of islands over a large 
