Whether the small variation near the north-east isles arose from any peculiar attraction, or from some oversight in taking the amplitude, I cannot determine; if from the latter, it would appear that the variation is a degree and a half less on the south-west, than on the east and north sides of Groote Eylandt. Scarcely any run of _tide_ was perceptible in North-west Bay, nor did the rise appear to exceed four or five feet at any part of the island, though it runs with some strength off the projecting points. The irregularity in different places was such, that the time of high water could not be ascertained; but I think there is only one full tide in the day, and that the flood comes from the northward. MONDAY 17 JANUARY 1803 Early on the 17th we worked out of the bay, and stretched off to sea with a W. N. W. wind; at noon the latitude was 13° 27' 10", and the furthest extreme of Chasm Island bore S. 26° W. After a calm in the afternoon, the sea breeze came in, and we steered south-westward till nine o'clock; when a bower anchor was let go in 14 fathoms, two or three miles from the north end of Winchilsea Island. In the morning [TUESDAY 18 JANUARY 1803] we lay up south-west, on the starbord tack, and weathered the island, leaving a rock one mile and a half on the other side. I wished, by a good bearing of Connexion Island, to join the survey completely round Groote Eylandt; and at nine o'clock it was set at S. 27½° to 47° W., two leagues. The wind then came ahead, and we tacked towards two small isles, where the anchor was dropped at ten, one mile and a half from their south side, in 16 fathoms, sand and shells. Our latitude here was 13° 43' 42" south, and the east side of Connexion Island bore S. 9½° W. six or seven miles; the difference of longitude from our situation on the 5th at three p.m., was hence ascertained to be it 1' 55" east, not differing 5" from what was given by No. 543, but No. 520 showed 6½' too much; the differences of longitude by the former time keeper alone have therefore been used round Groote Eylandt.