The gem stratum called mellan is always well defined, and occurs at a certain depth, which seems to correspond to the bottom of the lake at a definite period. This depth varies from two to twenty feet, and is perhaps even greater; but the natives rarely excavate below the depth of twenty feet. This peculiar formation, which is generally horizontal, is composed of a conglomerate of quartz gravel resting upon or mixed with a stiff clay, often indurated by a ferruginous oxide. In among this cascalho, or just below it and adhering to it, are found the fine pebbles and crystals of sapphire, tourmaline, garnet, zircon, spinel, and chrysoberyl. Under these rocks and in peculiar hollows in the plastic clay, which the natives call elephants’ footsteps, the gems are found clustered together heterogeneously, and often so perfect in form as to appear as though created there. At other places they are collected together in these pockets in such a manner as to suggest the idea that they had been washed in by a current of water. All these varieties of gems, some of them widely differing from each other in composition and form of crystallization, are here embedded together, and seem to have one common origin. This is the true matrix, and the gems are not found in other portions of the soil unless some disturbing force has removed them, like a strong current of water breaking up the cascalho and transporting the gems to alluvions of its own deposit. It is maintained and generally believed by mineralogists that the sapphire is formed in crystalline rocks; that in process of time the matrix is disintegrated, the gems set free, and washed down to the alluvial soils where they are now found. It is also thought that the gem-seekers might with patient care trace the precious stones to their source in the primitive ledges or the mountains, as the gold miner often follows for long distances the particles of gold in the soil until he discovers the parent vein in the solid ledge. But in Ceylon this view is not entertained by the natives; and all scientific efforts to find the sapphires in the mountain ledges have utterly failed. All trace of the sapphire and its attendant gems ceases as soon as we reach the limit of the gem stratum, and what seems to have once formed the shore of the lagoon. Beyond this plainly marked outline we may search in vain for the least sign of a connection with the older rocks either adjoining or at a distance. The result is the same if we examine the ledges on the same level or those of a higher elevation.