The manner in which, according to our theory, the arrangement of species in groups is due to partial extinction, will perhaps be rendered clearer in the following way. Let us suppose in any one great class, for instance in the Mammalia, that every species and every variety, during each successive age, had sent down one unaltered descendant (either fossil or living) to the present time; we should then have had one enormous series, including by small gradations every known mammiferous form; and consequently the existence of groups{450}, or chasms in the series, which in some parts are in greater width, and in some of less, is solely due to former species, and whole groups of species, not having thus sent down descendants to the present time. {450} The author probably intended to write "groups separated by chasms." With respect to the "analogical" or "adaptive" resemblances between organic beings which are not really related{451}, I will only add, that probably the isolation of different groups of species is an important element in the production of such characters: thus we can easily see, in a large increasing island, or even a continent like Australia, stocked with only certain orders of the main classes, that the conditions would be highly favourable for species from these orders to become adapted to play parts in the economy of nature, which in other countries were performed by tribes especially adapted to such parts. We can understand how it might happen that an otter-like animal might have been formed in Australia by slow selection from the more carnivorous Marsupial types; thus we can understand that curious case in the southern hemisphere, where there are no auks (but many petrels), of a petrel{452} having been modified into the external general form so as to play the same office in nature with the auks of the northern hemisphere; although the habits and form of the petrels and auks are normally so wholly different. It follows, from our theory, that two orders must have descended from one common stock at an immensely remote epoch; and we can perceive when a species in either order, or in both, shows some affinity to the other order, why the affinity is usually generic and not particular--that is why the Bizcacha amongst Rodents, in the points in which it is related to the Marsupial, is related to the whole group{453}, and not particularly to the Phascolomys, which of all Marsupialia is related most to the Rodents. For the Bizcacha is related to the present Marsupialia, only from being related to their common parent-stock; and not to any one species in particular. And generally, it may be observed in the writings of most naturalists, that when an organism is described as intermediate between two _great_ groups, its relations are not to particular species of either group, but to both groups, as wholes. A little reflection will show how exceptions (as that of the Lepidosiren,