I remarked, in the third chapter, that a reef, growing on a detached bank, would tend to assume an atoll-like structure; if, therefore, corals were to grow up from a bank, with a level surface some fathoms submerged, having steep sides and being situated in a deep sea, a reef not to be distinguished from an atoll, might be formed: I believe some such exist in the West Indies. But a difficulty of the same kind with that affecting the crater theory, runners, as we shall presently see, this view inapplicable to the greater number of atolls.
No theory worthy of notice has been advanced to account for those barrier-reefs, which encircle islands of moderate dimensions. The great reef which fronts the coast of Australia has been supposed, but without any special facts, to rest on the edge of a submarine precipice, extending parallel to the shore. The origin of the third class or of fringing-reefs presents, I believe, scarcely any difficulty, and is simply consequent on the polypifers not growing up from great depths, and their not flourishing close to gently shelving beaches where the water is often turbid.
What cause, then, has given to atolls and barrier-reefs their characteristic forms? Let us see whether an important deduction will not follow from the consideration of these two circumstances, first, the reef-building corals flourishing only at limited depths; and secondly, the vastness of the areas interspersed with coral-reefs and coral-islets, none of which rise to a greater height above the level of the sea, than that attained by matter thrown up by the waves and winds. I do not make this latter statement vaguely; I have carefully sought for descriptions of every island in the intertropical seas; and my task has been in some degree abridged by a map of the Pacific, corrected in 1834 by MM. D'Urville and Lottin, in which the low islands are distinguished from the high ones (even from those much less than a hundred feet in height) by being written without a capital letter; I have detected a few errors in this map, respecting the height of some of the islands, which will be noticed in the Appendix, where I treat of coral formations in geographical order. To the Appendix, also, I must refer for a more particular account of the data on which the statements on the next page are grounded. I have ascertained, and chiefly from the writings of Cook, Kotzebue, Bellinghausen, Duperrey, Beechey, and Lutke, regarding the Pacific; and from Moresby (See also Captain Owen's and Lieutenant Wood's papers in the "Geographical Journal", on the Maldiva and Laccadive Archipelagoes. These officers particularly refer to the lowness of the islets; but I chiefly ground my assertion respecting these two groups, and the Chagos group, from information communicated to me by Captain Moresby.) with respect to the Indian Ocean, that in the following cases the term "low island" strictly means land of the height commonly attained by matter thrown up by the winds and the waves of an open sea. If we draw a line (the plan I have always adopted) joining the external atolls of that part of the Low Archipelago in which the islands are numerous, the figure will be a pointed ellipse (reaching from Hood to Lazaref Island), of which the longer axis is 840 geographical miles, and the shorter 420 miles; in this space (I find from Mr. Couthouy's pamphlet (page 58) that Aurora Island is about two hundred feet in height; it consists of coral-rock, and seems to have been formed by the elevation of an atoll. It lies north-east of Tahiti, close without the line bounding the space coloured dark blue in the map appended to this volume. Honden Island, which is situated in the extreme north-west part of the Low Archipelago, according to measurements made on board the "Beagle", whilst sailing by, is 114 feet from the SUMMIT OF THE TREES to the water's edge. This island appeared to resemble the other atolls of the group.) none of the innumerable islets united into great rings rise above the stated level. The Gilbert group is very narrow, and 300 miles in length.