ere followed by such a number of
the sooty petrels as we had never seen equalled. There was a stream of
from fifty to eighty yards in depth, and of three hundred yards, or more,
in breadth; the birds were not scattered, but flying as compactly as a
free movement of their wings seemed to allow; and during a full _hour and
a half_, this stream of petrels continued to pass without interruption,
at a rate little inferior to the swiftness of the pigeon. On the lowest
computation, I think the number could not have been less than a hundred
millions; and we were thence led to believe, that there must be, in the
large bight, one or more uninhabited islands of considerable size.*

[* Taking the stream to have been fifty yards deep by three hundred in
width, and that it moved at the rate of thirty miles an hour, and
allowing nine cubic yards of space to each bird, the number would amount
to 151,500,000. The burrows required to lodge this quantity of birds
would be 75,750,000; and allowing a square yard to each burrow, they
would cover something more than 181 geographic square miles of ground.]

From the north-east point of the three-hummock land, the shore trended W.
1 deg. N. three miles; then S. 39 deg. W. four miles, to a rocky point, forming
the south-west extremity of what was then ascertained to be THREE-HUMMOCK
ISLAND. The channel which separates it from the land to the west, is, at
least, two miles in width, and is deep; so that it was difficult to
conjecture how the Indians were able to get over to the island. It was
almost certain that they had no canoes at Port Dalrymple, nor any means
of reaching islands lying not more than two cables length from the shore;
and it therefore seemed improbable that they should possess canoes here.
The small size of Three-hummock Island rendered the idea of fixed
inhabitants inadmissible; and whichever way it was considered, the
presence of men there was a problem difficult to be resolved.*

[* Future visitants to these islands have seen the Indians passing over
in bodies, by swimming, similar to those whom Dampier saw on the
north-west coast of New Holland. Why the natives of Port Dalrymple should
not have had recourse to the same expedient, where the distance to be
traversed is so much less, seems incomprehensible.]

The coast on the west side of the channel lies nearly south, and rises in
height as it advances towards the cliffy head, set on the 6th p.m. The
north end of this land is