FORMATION OF CORAL RHEFS AND ISLANDS. 245 
of an opening through a reef may require, therefore, no other 
explanation ; and it is obvious that harbors may generally be 
expected to exist wherever the character of the coast is such 
as to produce currents and give a fixed direction to them. 
The currents, about the reef grounds west of the large 
Feejee Islands, aid in distributing the debris both of the land 
and the reefs. In some parts, the currents eddy and deposit 
their detritus; in others they sweep the bottom clean. Thus, 
under these varying conditions, there may be growing corals 
over the bottom in some places and not in others; and the 
reefs may be distributed in patches, when without such an 
influence we mjght expect a general continuity of coral reef 
over the whole reef-grounds. 
The results from marine currents are often increased by 
waters from the island streams; for the coves, where harbors 
are most likely to be found, are also the embouchures of val- 
leys and the streamlets they contain. The fresh waters poured 
in add to the amount of water, and increase the rapidity of 
the out-current. At Apia, Upolu, there is a stream thirty 
yards wide; and many other similar instances might be men- 
tioned. These waters from the land bring down also much 
detritus, especially during freshets, and the depositions aid 
those from marine currents in keeping the bottom clear of 
growing coral. These are the principal means by which fresh- 
water streams contribute toward determining the existence of 
harbors; for little is due to their freshening the salt waters of 
the sea. 
The small influence of the last-mentioned cause—the one 
most commonly appealed to—will be obvious, when we con- 
sider the size of the streams of the Pacific islands, and the 
fact that fresh water is lighter than salt, and therefore, in- 
stead of sinking, flows on over its surface. The deepest rivers 
