590 CORALS AND CORAL ISLANDS. 
great width. The substratum, however, is, in general, contin- 
uous coral-rock; and if these more elevated parts were re- 
moved by any process, after an elevation, they would leave a 
nearly level area of coral limestone often as extensive as the 
whole reef-grounds. In an island like Dean’s, one of the Pau- 
motus, these reef-grounds are 1,000 square miles in extent. 
Still greater are the Bahama banks, the largest of them being 
390 miles long and 200 wide, an intervening “tongue of the 
ocean’ excepted. 
But the most extensive reef-grounds of the oceans are after 
all of small breadth compared with many of the ancient lime 
stones of the continents; and the reef-rocks also are peculiar 
in their very abrupt limits, the margins sometimes descending 
at a steep angle a thousand feet or more. ‘These differences 
between the new and the old arise in part from the fact that 
the coral reefs of the present era are made about small oceanic 
lands, or along the edges of the continents, while the limestones 
of ancient time were generally formed over the broad surface 
of a continent as it lay slightly submerged. 4 
The Abrolhos reefs of the Brazilian coast, described on page 
140, illustrate one of the methods by which the coral banks ex- 
tend and finally coalesce into beds of wide extent; but these 
are small compared with the great limestones of early time, 
and owe their slight approximation to them as regards extent 
to the wide range of shallow waters there afforded. These 
Abrolhos reefs differ from most limestone beds also in being 
formed largely of the corals in the position of growth. 
The tendency of modern reefs to grow up to the surface in 
narrow banks, separated by channels, appears to be unlike any 
thing we discover in the old rocks; and it seems to be an un- 
avoidable result of growth in the sea, where the waves pile up 
barriers, and the currents make, and keep open, channels. The 
case of the Australian and Feejee reefs are good examples It 

