The manner in which fishes are enabled to migrate from lake to lake is unknown, but many suggestions have been made. It is a fact that whirlwinds and waterspouts sometimes carry living fish in considerable numbers and drop them on the land. Here is one mode which might certainly have acted now and then in the course of thousands of years, and the eggs of fishes may have been carried with even greater ease. Again we may well suppose that some of these fish have once inhabited the streams that enter or flow out of the lakes as well as the lakes themselves; and this opens a wide field for conjecture as to modes of migration, because we know that rivers have sometimes changed their courses to such an extent as to form a union with distinct river basins. This has been effected either by floods rising over low watersheds, by elevations of the land changing lines of drainage, or by ice blocking up valleys and compelling the streams to flow over watersheds to find an outlet. This is known to have occurred during the glacial epoch, and is especially manifest in the case of the Parallel Roads of Glenroy, and it probably affords the true solution of many of the cases in which existing species of fish inhabit distinct river basins whether in streams or lakes. If a fish thus wandered out of one river-basin into another, it might then retire up the streams to some of the lakes, where alone it might find conditions favourable to it. By a combination of the modes of migration here indicated it is not difficult to understand how so many species are now common to the lakes of Wales, Cumberland, and Scotland, while others less able to adapt themselves to different conditions have survived only in one or two lakes in a single district; or these last may have been originally identical with other forms, but have become modified by the particular conditions of the lake in which they have found themselves isolated.