Referring chiefly, but not exclusively (from difficulty of transport, fewness, and the distinct characteristics of groups) to Mammalia; and first considering the three or four main [regions] divisions; North America, Europe, Asia, including greater part of E. Indian Archipelago and Africa are intimately allied. Africa most distinct, especially most southern parts. And the Arctic regions, which unite N. America, Asia and Europe, only separated (if we travel one way by Behring's St.) by a narrow strait, is most intimately allied, indeed forms but one restricted group. Next comes S. America,--then Australia, Madagascar (and some small islands which stand very remote from the land). Looking at these main divisions separately, the organisms vary according to changes in condition{122} of different parts. But besides this, barriers of every kind seem to separate regions in a greater degree than proportionally to the difference of climates on each side. Thus great chains of mountains, spaces of sea between islands and continents, even great rivers and deserts. In fact the amount <of> difference in the organisms bears a certain, but not invariable relation to the amount of physical difficulties to transit{123}. {122} In the _Origin_, Ed. i. p. 346, vi. p. 493, the author begins his discussion on geographical distribution by minimising the effect of physical conditions. He lays great stress on the effect of _barriers_, as in the present Essay. {123} Note in the original, "Would it be more striking if we took animals, take Rhinoceros, and study their habitats?" There are some curious exceptions, namely, similarity of fauna of mountains of Europe and N. America and Lapland. Other cases just <the> reverse, mountains of eastern S. America, Altai <?>, S. India <?>{124}: mountain summits of islands often eminently peculiar. Fauna generally of some islands, even when close, very dissimilar, in others very similar. [I am here led to observe one or more centres of creation{125}.] {124} Note by Mr A. R. Wallace. "The want of similarity referred to, is, between the mountains of Brazil and Guiana and those of the Andes. Also those of the Indian peninsula as compared with the Himalayas. In both cases there is continuous intervening land. "The islands referred to were, no doubt, the Galapagos for dissimilarity from S. America; our own Islands as compared with Europe, and perhaps Java, for similarity with continental Asia." {125} The arguments against multiple centres of creation are given in the _Origin_, Ed. i. p. 352, vi. p. 499.