We tacked to the westward in the afternoon, and an island came in sight, lying to the north of the two mounts, with several rocks and islets scattered on its north-east side. At sunset the wind died away, and a stream anchor was dropped in 16 fathoms sandy ground; our situation being five miles from the shore under Mount Dundas, and three from the nearest rocky islets to the north-west. The flood tide set gently to the westward, and induced me to suppose there might be a passage within the island and rocks, and in the morning [SATURDAY 12 FEBRUARY 1803] our endeavours were used to reach it; but the winds being light and mostly contrary, the evening came before we got through. An anchor was then dropped in 4 fathoms, coarse sand, one mile and a half from the sandy shore under Mount Saunders, and three miles from the south-west end of the island. The passage is more than two miles wide, and our soundings in working through it were between 4½ and 6 fathoms on a gravelly bottom; but afterwards we had little more in some places than 3 fathoms. [NORTH COAST. MELVILLE BAY.] Two natives, with a canoe, had been seen upon the island; and as our boat stood that way, sounding ahead of the ship, they waved and called to the people. The island is about five miles long, and between one and two in breadth; it is low, mostly destitute of wood, and the shores in general are sandy; and not being laid down in the Dutch chart, I distinguish it, with the islets and rocks to the north and north-east, by the name of _Melville Isles_: the south end which forms the passage, lies in 12° 8½' south, and 136° 52' east. In the opposite shore, between Mount Saunders and Dundas, is a sandy bight where ships would be sheltered from all winds except those at north-east, if the water be deep enough for them. The trees upon the hills showed a dark-green foliage; but the low land, especially under Mount Saunders, was sandy and barren, and so continued for seven miles westward, to a low point near a woody islet. Further on, the coast took a northern direction, and was seen from the mast head as far as N. N. W.; but no other part could be set from the deck than the highest of several eminences on the back land, named _Mount Bonner_, which proved to be an useful mark in the survey. The bearings taken at this anchorage were principally these: Mount Dundas, bluff north end, S. 54° E. Woody islet, near a western sandy point, S. 62 W. Mount Bonner, N. 82 W. Melville Isles, the northernmost, N. 13 E. Melville Isles, the largest, N. 83° E. to East. SUNDAY 13 FEBRUARY 1803