four and a half fathoms; but the great width of the reef, stretching three miles and a half southward of the land (which is represented in the drawing in the “Atlas of the _ Coquille’s_ Voyage” as descending abruptly to the water) shows, on the principle explained in the beginning of the last chapter, that it belongs to the barrier class. I may here mention, from information communicated to me by the Rev. W. Ellis, that on the N.E. side of _Huaheine_ there is a bank of sand, about a quarter of a mile wide, extending parallel to the shore, and separated from it by an extensive and deep lagoon; this bank of sand rests on coral-rock, and undoubtedly was originally a living reef. North of Bolabola lies the atoll of _Toubai_ (Motou-iti of the “_Coquille’s_ Atlas”) which is coloured dark blue; the other islands, surrounded by barrier-reefs, are pale blue; three of them are represented in Figs 3, 4, and 5, in Plate I. There are three low coral-groups lying a little E. of the Society Archipelago, and almost forming part of it, namely _ Bellinghausen_, which is said by Kotzebue (“Second Voyage,” vol. ii, p. 255), to be a lagoon-island; _Mopeha_, which, from Cook’s description (“Second Voyage,” book iii, chap. i), no doubt is an atoll; and the _Scilly_ Islands, which are said by Wallis (“Voyage,” chap. ix) to form a _group_ of _low_ islets and shoals, and, therefore, probably, they compose an atoll: the two former have been coloured blue, but not the latter. MENDANA OR MARQUESAS Group.—These islands are entirely without reefs, as may be seen in Krusenstern’s Atlas, making a remarkable contrast with the adjacent group of the Society Islands. Mr. F. D. Bennett has given some account of this group, in the seventh volume of the _Geograph. Journ._ He informs me that all the islands have the same general character, and that the water is very deep close to their shores. He visited three of them, namely, _Dominicana, Christiana,_ and _Roapoa_; their beaches are strewed with rounded masses of coral, and although no regular reefs exist, yet the shore is in many places lined by coral-rock, so that a boat grounds on this formation. Hence these islands ought probably to come within the class of fringed islands and be coloured red; but as I am determined to err on the cautious side, I have left them uncoloured.