138 
THEORY OF THE FORMATION 
Ch. V. 
would be the necessary result ; and it would be sepa- 
rated from the main land by a deep-water channel, 
broad in proportion to the amount of subsidence, and 
to the less or greater inclination of the bed of the sea. 
The effect of the continued subsidence of a barrier- 
reef, and its probable conversion into a chain of 
separate atolls, will be considered when we discuss the 
progressive disseverment of the larger Maldiva atolls. 
We now are able to perceive that the close similarity 
in form, dimensions, structure, and relative position 
between fringing and encircling barrier-reefs, and be- 
tween these latter reefs and atolls, is the necessary result 
of the transformation, during subsidence, of the one 
class into the other. On this view, the three classes of 
reefs ought to graduate into each other. Eeefs having 
an intermediate character between those of the fring- 
ing and barrier classes do exist ; for instance, on the 
S.W. coast of Madagascar, a reef extends for several 
miles, within which there is a broad channel from 
7 to 8 fathoms deep, but the sea does not deepen 
abruptly outside the reef. Such cases, however, are 
open to doubt, for an old fringing-reef which had 
extended itself on a basis of its own formation, would 
hardly be distinguishable from a barrier-reef produced 
by a small amount of subsidence, and with its lagoon- 
channel nearly filled up with sediment during a long 
stationary period. Between barrier-reefs, encircling 
either a single lofty island or several small low ones, 
and atolls including a mere expanse of water, a striking 
series can be shown : and in proof of this, I need only 
