We have next to consider the methods by which past geographical conditions may be ascertained, a task which, though beset with difficulties, is very far from being a hopeless undertaking. As has already been pointed out, it is perfectly possible for the geologist to determine the circumstances of formation of the various kinds of rocks, to distinguish terrestrial from aquatic accumulations and, among the latter, to identify those which were laid down in the sea and those which were formed in some other body of water. By platting on a map all the marine rocks of a given geological date, an approximate estimate may be formed as to the extension of the sea over the present land for that particular epoch. It is obvious, however, that for those areas which then were land and now are covered by the sea, no such direct evidence can be obtained, and only indirect means of ascertaining the former land-connections can be employed. It is in the treatment of this indirect evidence that the greatest differences of opinion arise and, if two maps of the same continent for the same epoch, by separate authors, be compared, it will be seen that the greatest discrepancies between them are concerning former land-connections and extensions. The only kind of indirect evidence bearing upon ancient land-connections, now broken by the sea, that need be considered here is that derived from the study of animals and plants, both recent and fossil. All-important in this connection is the principle that the same or closely similar species do not arise independently in areas between which there is no connection. It is not impossible that such an independent origin of organisms which the naturalist would class as belonging to the same species may have occasionally taken place, but, if so, it must be the rare exception to the normal process. This principle leads necessarily to the conclusion that the more recently and broadly two land-areas, now separated by the sea, have been connected, the more nearly alike will be their animals and plants. Such islands as Great Britain, Sumatra and Java must have been connected with the adjacent mainland within a geologically recent period, while the extreme zoölogical peculiarity of Australia can be explained only on the assumption that its present isolation is of very long standing. The principle applies to the case of fossils as well as to that of modern animals, and has already been made use of, in a preceding section, in dealing with the ancient land-connections of North America. It was there shown that the connection of this continent with the Old World and the interruptions of that connection are reflected and recorded in the greater or less degree of likeness in the fossil mammals at any particular epoch. Conversely, the very radical differences between the fossil mammals of the two Americas imply a long-continued separation of those two continents, and their junction in the latter half of the