It is from this epoch that, in all probability, we should date the commencement of what may be termed "the volcanic history" of the moon. We must bear in mind that although the moon's surface had become solid, its temperature may have remained high for a very long period. But the continuous radiation of the surface-heat into space would produce continuous contraction, while the convection of the interior heat would tend to increase the thickness of the outer solid shell; and this, ever pressing with increasing force on the interior molten mass, would result in frequent ruptures of the shell, and the extrusion of molten lava rising from below. Hence we may suppose the fissure-eruptions of lava were of frequent occurrence for a lengthened period during the early stage of consolidation of the lunar crust; but afterwards these may be supposed to have given place to eruptions through pipes or vents, resulting in the formation of the circular craters which form such striking and characteristic objects in the physical aspect of our satellite.[14] It is not to be supposed that the various physical features on the lunar surface have all originated in the same way. The great ranges of mountains previously described may have originated by a process of piling up of immense masses of molten lava extruded from the interior through vents or fissures; while the great hollows (or "seas") are probably due to the falling inwards of large spaces owing to the escape of the interior lava. But it is with the circular craters that we are most concerned. Judging from analogy with the lava-craters present on our globe, we must suppose them to be due to the extrusion, and piling up, of lava through central pipes, followed in some cases by the subsidence of the floor of the crater. It seems not improbable that it was in this way the greater number of the circular craters lying around Tycho, and dotting so large a space round the margin of the moon, were constructed. (See Fig. 38.) In general they appear to consist of an elevated rim, enclosing a depressed plain, out of which a central cone arises. The rim may be supposed to have been piled up by successive discharges of lava from a central orifice; and after the subsidence of the paroxysm the lava still in a molten condition may have sunk down, forming a seething lake within the vast circular rampart, as in the case of the Hawaiian volcanoes. The terraces observable within the craters in some instances have probably been left by subsequent eruptions which have not attained to the level of preceding ones; and where a central crater-cone is seen to rise within the caldron, we may suppose this to have been built up by a later series of eruptions of lava through the original pipe after the consolidation of the interior sea of lava. The mamelons of the Isle of Bourbon,[15] and some of the lava-cones of Hawaii, appear to offer examples on our earth's surface of these peculiar forms.