Between the tropics, islands are formed from the mere accumulation of coral; and it is the peculiarity of those regions, to produce rocks that have not passed through the usual process of mineral consolidation.[208] The islets, however, which are thus formed, must have their bases laid on a solid rock, though perhaps at a great depth; and it is not probable, that after they are once raised above the surface of the sea, they can still rise farther, except by some elevation of the rock which serves as their foundation.[209] Now, at Palmerston island, which comprehends nine or ten low islets, that may be reckoned the heads of a great reef of coral rock, Captain Cook informs us of his having seen, "far beyond the reach of the sea, even in the most violent storms, elevated coral rocks, which, on examination, appeared to have been perforated in the same manner that the rocks are that now compose the outer edge of the reef. This evidently shows," he adds, "that the sea had formerly reached so far; and some of these perforated rocks were almost in the centre of the island."[210] [Footnote 208: Dr Foster, in his Voyage round the World, (vol. ii. p. 146,) gives an instance in the South Sea Islands, where the surface of the island, though entirely a coral rock, was raised forty feet above the level of the sea.] [Footnote 209: A very curious account of the formation of such islands is given by A. Dalrymple, Esq., in the Philosophical Transitions, vol. lvii. p. 394.] [Footnote 210: Cook's Third Voyage, vol i. p. 221.] The same excellent navigator, giving an account of the peninsula at Cape Denbigh, remarks: "It appeared to me, that this peninsula must have been an island in remote times; for there were marks of the sea having flowed over the isthmus." 402. We are here touching on one of those subjects, where we feel much the want of accurate and ancient observations, and where it is not from the infancy, but the maturity of science that any thing approaching to certainty can be looked for. The utmost that we can expect at present, is an anticipation, which future ages must certainly modify and correct. The best thing, in the mean time, that can be done for the advancement of this branch of geological knowledge, is to ascertain with exactness the relative level of the sea, and of such points upon the land as can be distinctly marked, and pointed out to succeeding ages. This is not so easy as it may at first appear. Where every object changes, it is difficult to find a measure of change, or a fixed point from which the computation may begin. The astronomers already feel this inconvenience, and when they would refer their observations to an immoveable plane, that shall preserve its position the same in all ages, they meet with difficulties, which cannot be removed but by a profound mathematical investigation.