144 
THEORY OF THE FORMATION 
Cn V. 
divides Mahlos Mahdoo (Plate II., fig. 4) would almost 
necessarily be formed. The scattered reefs situated 
near the borders of the new channel, from being favour- 
ably placed for the growth of coral, would, by their 
extension, tend to produce fresh margins to the dis- 
severed portions: and a tendency of this kind is evident 
in the elongated reefs which border the two channels 
intersecting Mahlos Mahdoo. Such channels would 
become deeper with continued subsidence, and, from 
the reefs on both sides not growing up perpendicularly, 
somewhat broader. In this case, and more especially 
if the channels had been originally formed of consider- 
able breadth, the dissevered portions would soon be- 
come perfect and distinct atolls like Ari and Ross atolls 
(Plate II., fig. 6), or like the two Nillandoo atolls, which 
must be considered as distinct, although plainly related 
to each other in form and position, and separated only 
by moderately deep channels. Further subsidence 
would render such channels unfathomable, and the 
dissevered portions would then resemble Phaleedoo and 
Moluque atolls, or Mahlos Mahdoo and Horsburgh atolls 
(Plate II., fig. 4), which are related to each other only 
in proximity and position. Hence, on the theory of 
subsidence, the disseverment of large atolls which are 
exposed to strong currents and which have imperfect 
margins (for otherwise their disseverment would be 
scarcely possible) is far from being an improbable 
event ; and the several stages, from a close connection 
to the entire isolation of some of the atolls in the 
Maldiva Archipelago, are readily explicable. 
