Feb. 22, 1858.] 
WALLACE ON THE ARRU ISLANDS. 
165 
the east coast, are an immense number, extending to the extreme 
south, but nowhere reaching more than 15 or 20 miles from the 
central island. All are contained in a very shallow sea full of coral, 
and producing the pearl shells, which form the principal article of 
commerce in the islands. The whole of the islands are covered 
with a dense and very lofty forest. 
The physical features here described are of the greatest interest, 
and probably altogether unique, for I have been unable to call to 
mind any other islands in the world which are completely divided 
by salt-water channels, having the dimensions and every other 
character of true rivers. What is the real nature of these, and how 
they have originated, are questions which have occupied much of 
my attention, and which I have at length succeeded in answering, 
to my own satisfaction at least. There are three distinct modes by 
which islands may have been formed, or have arrived at their pre- 
sent condition, — elevation, subsidence, and separation from a conti- 
nent or larger island. Most volcanic islands have been elevated ; 
coral islands with lagoons or with barrier reefs have suffered sub- 
sidence ; while our own islands, Sicily, Ceylon, and many others, 
have no doubt been separated from the adjacent continents. Now, 
the Arru islands, being all coral rock, and the adjacent sea all 
shallow and full of coral, it would seem easy to account for their 
origin by supposing them to have been elevated gradually from 
beneath the water, as the much more lofty islands of Ke, sixty miles 
to the westward, have no doubt been. But in this case it is impos- 
sible to explain the formation of those regular river-like channels 
which cut across the largest and most elevated mass. A fissure 
produced during elevation will not explain it, for it has all the 
regular curves and windings of a river ; and the action of tides and 
currents combining with the elevating force will, indeed, well explain 
the origin of separate islands divided by channels, of varying width 
and depth, but cannot be imagined to have produced a true river- 
bed 40 miles in length and of the greatest regularity both in width 
and depth. If we suppose the subsidence of a more extensive 
island to have brought Arru to its present form, we shall find it 
equally difficult to account for these rivers, because the subsidence 
of any country with an irregular and undulating surface must, by 
allowing the sea to overflow all the level tracts, produce a most 
irregular distribution of water in the channels separating islands, 
and form deep inlets, creeks, and inland lochs, all of which are here 
absent. The only other way of accounting for the origin of the 
Arru Islands is, by supposing them to have once formed a part of 
the main land of New Guinea, from which they have been separated 
