In doing so it may be desirable, in the first place, to determine what form of crater on our earth's surface those of the moon do not represent; and we are guided in our inquiry by the consideration of the absence of water on the lunar surface. Now there are large numbers of crateriform mountains on our globe in the formation of which water has played an important, indeed essential, part. As we have already seen, water, though not the ultimate cause of volcanic eruptions, has been the chief agent, when in the form of steam at high pressure, in producing the explosions which accompany these eruptions, and in tearing up and hurling into the air the masses of rock, scoriæ, and ashes, which are piled around the vents of eruption in the form of craters during periods of activity. To this class of craters those of Etna, Vesuvius, and Auvergne belong. These mountains and conical hills (the domes excepted) are all built up of accumulations of fragmental material, with occasional sheets and dykes of lava intervening; and where eruptions have taken place in recent times, observation has shown that they are accompanied by outbursts of vast quantities of aqueous vapour, which has been the chief agent (along with various gases) in piling up the circular walls of the crater. It has also been shown that in many instances these crater-walls have been breached on one side, and that streams of molten lava which once occupied the cup to a greater or less height, have poured down the mountain side. Hence the form or outline of many of these fragmental craters is crescent-shaped. Such breached craters are to be found in all parts of the world, and are not confined to any one district, or even continent, so that they may be considered as characteristic of the class of volcanic crater-cones to which I am now referring. In the case of the moon, however, we fail to observe any decided instances of breached craters, with lava-streams, such as those I have described.[13] In nearly all cases the ramparts appear to extend continuously round the enclosed depression, solid and unbroken; or at least with no large gap occupying a very considerable section of the circumference. (See Fig. 38.) Hence we are led to suspect that there is some essential distinction between the craters on the surface of the moon and the greater number of those on the surface of our earth. It is scarcely necessary to add that the volcanic mountains of the moon offer no resemblance whatever to the dome-shaped volcanic mountains of our globe. If it were otherwise, the lunar mountains would appear as simple luminous points rising from a dark floor, over which they would cast a conical shadow. But the form of the lunar volcanic mountains is essentially different; as already observed, they consist in general of a circular rampart enclosing a depressed floor, sometimes terraced as in the case of Copernicus, from which rise one or more conical mountains, which are in effect the later vents of eruption.