166 
WALLACE ON THE ARRU ISLANDS. 
[Feb. 22 , 1858 . 
by the subsidence of an intervening district. The principal objec- 
tion to this view is the great width of open sea (from 100 to 200 
miles) between their eastern limits and the south-west coast of 
New Guinea. It is, however, to be observed, that this sea nowhere 
exceeds a depth of 40 fathoms, while immediately to the north, a 
fathomless sea reaches close up to the New Guinea coast, and also 
within 20 miles of Arru on the west. By supposing the central 
land of Arru to have remained unmoved during the subsidence, the 
present transverse channels may be explained as being in fact por- 
tions of actual rivers which flowed from the great central mountain- 
range of New Guinea, and here had their outlet after a course of 
two or three hundred miles. The position and direction of the Uta- 
nata and Wakua rivers in New Guinea, renders it not improbable 
that the Arru rivers may have been once the ccntinuation of them. 
In no other manner does it seem to me possible to explain the origin 
of these channels ; for I believe no example exists of anything but 
true rivers producing narrow, winding channels of regular width 
and depth through an undulating rocky country. If, therefore, there 
is only one cause in existing nature adequate to produce the effects 
visible, we must impute them to that cause, even though implying 
changes of sea and land of such an extensive character. 
We have, however, other evidence of a totally distinct nature, 
which gives a powerful support to this view of the origin of the 
Arru Islands. The distribution of the animals of Arru and New 
Guinea proves the close connection between these countries, it 
being evident that, where a considerable number of animals which 
have no means of passing from the one to the other are common to 
two countries, some former communication must have existed be- 
tween them. A few such cases of community may indeed be 
explained by the various accidents by which animals may be trans- 
ported from one country to another ; but when the community is 
more general, there is no such easy way of accounting for it. In 
the present case birds being almost the only animal productions of 
New Guinea of which anything is known, the argument must be 
drawn almost entirely from that class, which, it may be objected, 
can furnish no certain data, as they have the means of passing from 
one country to the other. It is, however, well known that birds 
have their geographical limits as accurately defined as other animals, 
and that many extensive groups are quite as unable to pass wide 
tracts of ocean as any quadrupeds can be. 
The first fact, then, is, that out of the small number of land birds 
known from all parts of the coast of New Guinea, or about one 
hundred, I have myself found thirty-six in Arru. This renders it 
