[[p. 193]] The editor having kindly sent me a proof of Mr. Jukes-Browne's paper, I beg to make a few remarks thereon.

     I cannot but think that Mr. Jukes-Browne's criticism of the Darwinian classification of islands, which I have adopted and more fully developed, is rather one of words and definitions than of realities. The very terms of the classification--"Oceanic" and "Continental"--show that it is a broad and wide-reaching one; and its main implication, the permanence of oceanic and continental areas, is equally broad and fundamental. That there should be islands situated upon the ever-fluctuating margin of these two areas which are difficult to class, or which may, at different geological periods, have possessed the characteristics of "oceanic" or of "continental" islands, is what might certainly be expected; the wonder is that there are so very few of them. Barbados is, technically, an oceanic island; but it is [[p. 194]] situated upon the old sea-margin of the American continent, and a portion of that old continental margin forms the base of its oceanic deposits. I recognised the possibility of such a base for some apparently oceanic islands in the passage quoted by Mr. Jukes-Browne, but was not then aware that any such existed. Of course, if old stratified rocks could be shown to form the base of any of the mid-oceanic islands, the whole classification, and the theory which is founded on it, might be imperilled; but this has not yet been done.

     It is evident that, with island groups whose components vary in size from many thousands of square miles to small sea-washed rocks, all definitions must be taken broadly and as applying to the group. Even among the pre-eminently continental British Isles there are many hundreds--out of the thousand of which they are said to consist--which have neither mammalian, amphibian, nor reptilian inhabitants; but it will hardly be objected that such cases as these upset the biological definition of continental islands. In the same manner the Seychelles are classed as belonging to the Madagascar group, and are, therefore, ancient continental, while Mauritius, Bourbon, and Rodriguez are true oceanic islands.

     I do not know why Mr. Jukes-Browne should say that I regard New Caledonia as an oceanic island. At p. 473 of Island Life I refer to it as probably once connected with New Zealand; and again, at p. 485, I suppose it to have once formed an extension of New Zealand, which, though in some respects anomalous, has all the main characteristics of a continental island.

     Looking at the question broadly, as a generalisation applying to all the well-marked islands and island-groups of the globe, I entirely deny the validity of the conclusions expressed in the last three paragraphs of Mr. Jukes-Browne's paper, conclusions which are founded exclusively on islands situated upon the margin of the continental area.