The characters of igneous rocks vary considerably according as they have consolidated under atmospheric pressure only, or under that of superincumbent rocks. We must remember also that submarine lavas have to sustain a pressure of an extra atmosphere for every thirty feet of depth, or 400 atmospheres at 2000 fathoms, and that such rocks have a claim to be regarded as deep-seated. The gases that igneous rocks contain, probably as essential features of the molten magma, and at a temperature above their critical points, escape to a large extent near or at the surface of the earth. The bubbles raised in lava, whereby it is rendered _scoriaceous_, and the clouds of vapour rising from cooling lava-flows and from the throat of a volcano in eruption, are sufficient evidences of this process. The extremely liquid lavas of Kilauea in Hawaii, which emit very little vapour, are notable as exceptions. In the case of masses that cool underground, the retention of gases, and ultimately of liquids, until a very late stage of consolidation retards crystallisation until temperatures are reached lower than those at which it starts in surface-flows. As A. Harker points out[60], "the loss of these substances, by raising the melting-points in the magma, may be the immediate cause of crystallisation, quite as much as any actual cooling." The formation of crystals in lavas is rapid, and the average crystals are therefore small, and often felted together in a mesh, the interstices of which are filled by residual glass. Slowness of cooling is the really important factor that affects the size of crystals, that is, the coarseness of grain, in igneous rocks. Pressure may promote crystallisation, by raising the melting-points of minerals; but, after a certain maximum effect in this direction, it is quite possible that an increase of pressure may actually lower the melting-points, and cause one or other mineral to remain in solution in the magma. It is not clear how pressure can affect the size of any constituent, except by bringing about conditions under which it can go on growing, while other constituents remain in solution, or do not grow so fast. Such conditions may arise from the aid given by pressure to the retention of what French geologists have called _agents minéralisateurs_. Several familiar minerals, for instance albite, orthoclase, and quartz, require the presence of water for their formation. Volatile substances, not utilised in the ultimate product, no doubt similarly assist the formation of many rock-forming minerals. Occasionally, moreover, as in the development of the micas and certain of the silicates known as zeolites, some proportion of hydrogen is retained by minerals thus crystallising from the magma. Micas appear to require the presence of fluorine for their development. J. P. Iddings[61], however, lays stress in this case on the chemical activity of hydrogen at high temperatures.