On Coral Reefs and Islands. 37 
none of these so-called rivers are over three feet in depth; and 
the most that they can do is to produce a thin layer of brackish 
bea over the sea within the channels. 
The annexed figure of the harbor of 
Falifa, Upolu, represents another coral harbor, 
as surveyed by Lieutenant Emmons. At its 
head there is a fine stream twenty-five or 
thirty yards wide, and three feet deep. Not- 
withstanding the unusual size of the river, 
the coral reef lies near its mouth, and projects 
some distance in front of it. Its surface is dead, 
but corals are —— upon its outer slope, 
f. The harbor of Rewa, in the F'eejees, may be again — 
to. The waters received by the bay amount to at least 50( 
cubic feet a minute. Yet there is an extensive reef sro 
the bay, lying but three miles from the shores, and with only. 
two narrow openings for ships. The case is so remarkable that 
we can hardly account for the facts without supposing the river’s 
mouth to have neared the reef by depositions of detritus since 
the inner parts of the reef were formed; and there is some evi- 
dence that this was the case, though et what distance, we can- 
not definitely state. With this admission) the facts may still sur- - 
prise us; yet they are explained on the principle t that fresh water 
does not sink i in the ocean, but is superficial, and runs on in a 
distinct channel ; its effect is almost wholly through hydrostatic 
pressure, and detritus ee Beside es freee potanbes, there 
HARBOR OF FALIFA. 
to the medias sand or earth of marine teal Sentidinidas cur- 
rents, in preventing the growth of coral, but little is due to the 
freshening influence of the streams of the islands. 
But while observing that currents have so decided an influence 
on the condition of harbors, we should remember another preva~ 
cause, already remarked upon, and perhaps more wide in its 
effects than those just considered. I refer to the features of the 
ha i i land, or the character of soundings off a coast. We 
t here what has already been a — 
that many oa the interruptions of reefs have thus ari The | 
ide break off Matavai may be of this kind. The pctv 
of the inner channel at Papieti, forming a space for —— 
ere fie: tooo eee E 
as 
lowed istance the shores, 
more — seppscckss: this part 
