At daylight of the 21st, having a fresh breeze at N. W., we prepared to depart, and hove short; but the ship driving before the sails were loosed, and there being little room astern, a second bower was dropped and a kedge anchor carried out. This last not holding after the bowers were weighed, a stream anchor was let go; and before the ship brought up, it was again necessary to drop the best bower. At this time we were not more than a cable's length from the rocks of Middle Island; and the ship being exposed to great danger with the least increase of wind, we got a spring on the stream cable and began to heave on the best bower. In the mean time the ship drove with both anchors ahead, which obliged me, on the instant, to cut both cables, heave upon the spring, and run up the jib and stay-sails; and my orders being obeyed with an alacrity not to be exceeded, we happily cleared the rocks by a few fathoms, and at noon made sail to the eastward. This example proved the anchorage in the eastern part of Goose-Island Bay to be very bad, the sand being so loose as not to hold the ship with two anchors, though the water was smooth and the wind not more than a double-reefed-top-sail breeze; yet further westward, between Goose Island and the west beach, our anchor had held very well before. The most secure situation should seem to be off the east end of the middle beach, between it and the rock, in 4 or 5 fathoms; but I cannot answer for the ground there being good, though to all appearance it should be the best in the bay. The _latitude_ observed from an artificial horizon on the middle beach was 34° 5' 23" south; and the _longitude_ of the place of observation, a little east of that before fixed by the time keepers from King George's Sound, (Vol. I.), will be 123° 9' 37.6" east. Mr. Flinders took three sets of altitudes between the 18th p.m. and 21st a.m., from which the rates of the time keepers, and their errors from Greenwich time at noon there of the 21st, were found to be as under; Earnshaw's No. 543, slow 3h 10' 59.53" and losing 19.63" per day. Earnshaw's No. 520, fast 1h 31' 54.28" and losing 34.07" per day. At the first observation, the longitudes deduced from the Coepang rates were, by No. 543--123° 33' 37.5" east, No. 520--123 25 22.5 east; the mean of which is 19' 52.4" more than what I consider to be the true longitude, but on using rates equally accelerated from those at Coepang to what were found above, the error becomes reduced to 12' 11.6" east; which is the sum of the apparent irregularity of the time keepers from April 8 to May 18, or in 40.2 days. The corrections applied to the longitude during the last passage, are therefore what arise from the equal acceleration of the rates, and from the proportional part of the 12' 11.6" of irregularity; and when thus corrected, the time keepers did not appear to differ at Cape Leeuwin and Mount Gardner more than 1' from the longitude of the former year.