THE ORIGIN OF CORAL REEFS. fin) 
extent the surface-currents acted, and to discover the source of the 
sand. It was also necessary to observe what changes in the form 
and extent of the islands had occurred in the experience of the 
residents during the half-century of their occupation. 
The westerly equatorial drift or south-east trade current, striking 
the south-east angle of the atoll, there divides and sweeps around the 
coasts, the two branches meeting and forming an eddy off the north- 
west island, a spot where drift timbers are often detained and 
stranded after having been swept around half the circumference of 
the atoll. Advantage of this current is taken by the proprietor of 
the islands, who directs his men to mark any logs of valuable timber 
thrown up on the weather or south-east coast, and then to launch 
them again outside the breakers. In this way huge logs are trans- 
ported by the current to any particular island. If left alone, the logs, 
whether drifted around the north or south side of the atoll, arrive 
finally in the eddy off the north-west angle. This current finds its 
way into the lagoon through the several passages between the 
islands, its rate there varying usually from half a knot to two knots 
inthe hour. Only rarely is there any check to the inflow of water 
through the passages, as, for instance, during north-west gales. 
The current in these passages carries daily a large amount of 
sand into the lagoon. I discovered this accidentally whilst using 
the tow-net for catching the pelagic animals brought in by the 
current. The source of this sand is the weather edge of the reef 
on the outer side of the islands, where the breakers are unceasingly 
at work in keeping up the supply. After several measurements 
under varying conditions of current, tide, and depth, I estimated 
that during every day of ordinary weather at least 10 tons of sand 
are carried through the passages into the lagoon. During gales 
and cyclones this amount is greatly increased; and probably the 
estimate for an ordinary year would not be less than 5,000 tons. 
The bulk of this sand is deposited by the current near the inner 
mouths of the passages and on the margins of the lagoon, where it 
goes to extend the islands in the form of banks stretching into the 
lagoon. In this manner an island obtains a horse-shoe shape, just 
as the \/-shaped ridge is formed by placing a stake in a river-bed. 
The first stage is represented by an island with two sand-banks 
extending into the lagoon, one from each extremity. The second 
stage is that in which the island has attained a semi-crescentic shape 
by the encroachment of its vegetation on the newly-formed banks. 
In the course of time, when the vegetation of the island has en- 
