If we look first at the mode of formation of overground perennial ice, that is, of the ice of glaciers and of rock gorges; and then at the evidences of the mode of formation of underground perennial ice, in boulder heaps, wells and caves; we will soon see that the transitions between them are gentle in character and that there is nothing unnatural about the formation of the ice in glacières. _Glaciers._--Everyone now knows the main characteristics of glaciers. They are formed in parts of the earth where the land or the mountains reach to the region of perpetual snow. The snows fall from the sky, and accumulate into a snow cap, which by its own weight and by melting and regelation, gradually changes to ice. This, by the laws of gravitation, descends to lower levels, and in mountain valleys extends sometimes far below the snow line into the region of cultivated fields. These valley prolongations of the perpetual snow caps are the glaciers. The important point to notice here, is that the formation of glaciers is originally entirely due to the precipitation of moisture by cold in the upper portions; while the destruction of glaciers is due to the action of heat melting the ice in the lower portions, where they disappear in the shape of streams of running water. It is, therefore, not surprising that the greatest glaciers are found in the Arctic and Antarctic regions and in the highest mountain ranges; and that in the tropics glaciers are either wanting or exceedingly small. _Gorges and Troughs._--Gorges and gullies, where ice remains over, are a transitional form between glaciers and glacières. In many mountain ravines or canyons, the enduring snow consists principally of the avalanches which have fallen from the heights above during the winter and solidified in the bottom of the ravines. Freezing gorges proper, however, are not dependent on avalanches for their supply, but they receive the accretions to their ice directly from the winter snows. These fall into the gorge itself and by melting and regelation gradually solidify into a mass of ice which, when well sheltered against sun and wind, remains over sometimes till the following winter. By their mode of formation, therefore, it is evident that the ice in these gorges has some of the characteristics of glaciers; that it is due to the same prime causes as the ice of glaciers or the ice on ponds and rivers, namely the cold of winter; and in fact, it is not far wrong to consider these gorges as miniature glaciers.