throw more detritus into the main valley than the river in the latter can at once dispose of. Partial dams are thus produced, and large valley-lakes form above the obstructions, of which the Silser See and Silvaplana See in Upper Engadine are examples. 5. Eolian Basins. Another class of basins owe their origin to the action of the wind. Some are erosion-basins caused by the removal of loose, weathered rock-material. Professor Pumpelly seems to have been the first to recognise basins of this kind, which were observed by him in Mongolia. They have since been encountered in many other regions, as in Bahia, in Central Asia, and elsewhere. Sometimes these basins form temporary lakes, at other times the water remains more or less persistently. Some interesting examples have been described by Mr. G. K. Gilbert as occurring in Arkansas and elsewhere in the Great Plains of North America. Basins of this kind are naturally confined to relatively dry regions—to regions where the rocks and soils are not sufficiently protected by vegetation. Reference may also be made to the temporary or more persistent lakes which owe their origin to the unequal distribution of wind-blown accumulations, some account of which has already been given. 6. Rock-Fall Basins. Rock-falls and landslips not infrequently disturb local drainage, and may cause lakes to appear. Many small lakes of this class occur in the Alps and other mountain regions where the geological structures are weak and liable to collapse. 7. Glacial Basins. The basins coming under this head are essentially of two kinds. Some are hollows of excavation, others owe their origin to the unequal heaping up of glacial and fluvio-glacial deposits. It is not always possible, however, to distinguish sharply between the two. In many cases excavation and accumulation have alike been concerned in their formation. The glacial origin of both is at once suggested by the fact that they are confined to regions which yield other and independent evidence of former glacial action. We note further that their presence has no immediate or direct connection with the character of the rocks or with the geological structure of the tracts in which they lie. They occur in crystalline, igneous, and schistose rocks, and in sedimentary strata of all kinds and of all degrees of induration-conglomerate, sandstone, greywacke, clay-slate, shale, limestone, gravel, etc. They are not restricted to areas of folded, contorted, and fractured rocks, but appear with all their characteristic features equally well developed in places where the strata are gently undulating and approximately horizontal. The formerly glaciated areas of the earth's surface are pre-eminently the lake-lands of the world. We have only to look at a series of good maps to see that this is the case. Taking Europe as an example, we find that very few lakes occur in regions over which ice-sheets and glaciers have not at one time extended, the most notable of those lakes being the volcanic basins of Auvergne, the Eifel, and Central Italy. What non-glaciated region of our continent can show a lake-dappled surface like Finland ? Where in extraglacial tracts can we find anything to compare with the paysage morainique of North Germany and Russia? Precisely the same phenomena confront us in North America. How abundantly are lakes distributed over all the vast tract formerly occupied by the great inland ice! South of the glacial boundaries they are practically unknown. We note further that the vertical distribution of the class of lakes now under consideration is not less suggestive of their origin. Cirque-lakes and other high-level lakes are not confined to any one region, they occur in mountain-tracts all the world over, wherever these have formerly nourished glaciers. Low-lying valley-lakes like those of the Alps have, on the other hand, a much more restricted distribution. They abound in the mountains of temperate latitudes, where great valley-glaciers formerly existed, but they are looked for in vain in the mountains of the warmer zones, the lower reaches of whose valleys have never been glaciated. Again, in the northern tracts of Europe and North America glacial basins are not even confined to mountain-valleys, but occur more or less abundantly over the lowlands that sweep out from the mountains. In a word, there is a close connection between glaciation and the development of lake-basins. Basins of glacial origin naturally vary much in char acter, according to their position and the particular mode of their formation. Some, as mentioned above, are rock-basins, others are barrier-basins, and many are partly both. It must be added that not a few lakes met with in glaciated regions are not of glacial origin. This is particularly the case in mountain-valleys, where barrier-basins have often been formed by rock-falls and fluviatile action. Glacial basins may be roughly grouped as follows: 1. Cirque or Corrie basins. 2. Mountain-valley basins. 3. Lowland and Plateau basins. 5. Submarine basins. 1. Cirque or Corrie basins are confined to mountain regions. Frequently they appear as niche-like indentations on mountain-slopes at high elevations above the bottoms of the adjacent valleys. At other times they are set farther back from the brow of a valley, forming cup-shaped depressions in the flanks of the higher crests and ridges. When such is the case the water escaping from them may flow for a longer or shorter distance before it reaches the terminal shoulder of a mountain to plunge downwards to the valley below. In detail, cirques vary in character with the nature of the rocks and their geological structure. Many have a crater-like appearance, some of the wider ones resembling the section of a steepsided amphitheatre, while the narrower ones show more abrupt slopes. Although now and again the converging slopes may be relatively smooth and not so steep, yet as a rule they are rugged and precipitous, showing bare, gaunt walls of rock, trenched and furrowed by torrent action and shattered by frost. In regions which have formerly supported glaciers cirques are more or less flat-bottomed, or saucer-shaped, and consequently many are occupied by lakes. It is worthy of note that such corrie-lakes, or tarns, are usually deeper in proportion to their extent than the large valley-lakes of lower levels. Many corrie-lakes rest in true rock-basins; others seem to be wholly dammed by moraines; while yet others are partly rock-basins, partly barrier-basins. Not a few have been drained by the water escaping from them cutting back its channel. Others, again, would seem to be filled up by rock-falls and the detritus and débris shot down from the surrounding heights. Many cirques, on the other hand, have never contained lakes, their flat bottoms sloping gently, but continuously, outwards. That cirque-basins have been formerly occupied by glaciers is shown by the presence of moraines and the frequent appearance of roches moutonnées and striæ, the direction of which indicates an outflow of ice from the depressions. These marks of glacial action are confined to the bottom of a cirque; the precipitous rock-walls show none. The question of the origin of cirque-lakes has sometimes been obscured by confounding the origin of the cirques with that of the basins which occupy their |