The method we have followed in these investigations is to accept the results of geological and palaeontological science, and the ascertained facts as to the powers of dispersal of the various animal groups; to take full account of the laws of evolution as affecting distribution, {449} and of the various ocean depths as implying recent or remote union of islands with their adjacent continents; and the result is, that wherever we possess a sufficient knowledge of these various classes of evidence, we find it possible to give a connected and intelligible explanation of all the most striking peculiarities of the organic world. In Madagascar we have undoubtedly one of the most difficult of these problems; but we have, I think, fairly met and conquered most of its difficulties. The complexity of the organic relations of this island is due, partly to its having derived its animal forms from two distinct sources--from one continent through a direct land-connection, and from another by means of intervening islands now submerged; but, mainly to the fact of its having been separated from a continent which is now, zoologically, in a very different condition from that which prevailed at the time of the separation; and to its having been thus able to preserve a number of types which may date back to the Eocene, or even to the Cretaceous, period. Some of these types have become altogether extinct elsewhere; others have spread far and wide over the globe, and have survived only in a few remote countries--and especially in those which have been more or less secured by their isolated position from the incursions of the more highly-developed forms of later times. This explains why it is that the nearest allies of the Madagascar fauna and flora are now so often to be found in South America or Australia--countries in which low forms of mammalia and birds still largely prevail;--it being on account of the long-continued isolation of all these countries that similar forms (descendants of ancient types) are preserved in them. Had the numerous suggested continental extensions connecting these remote continents at various geological periods been realities, the result would have been that all these interesting archaic forms, all these defenceless insular types, would long ago have been exterminated, and one comparatively monotonous fauna have reigned over the whole earth. So far from explaining the anomalous facts, the alleged continental extensions, had they existed, would have left no such facts to be explained. * * * * * {450} CHAPTER XX ANOMALOUS ISLANDS: CELEBES