On Canna and Sanday the traces of a river are preserved which poured its flood-waters across the lava-fields in that part of the volcanic region, while streams of basalt were still from time to time issuing from vents and fissures. Not more than fourteen miles to south-east stands the Scuir of Eigg, with its buried river-channel and its striking evidence that there, also, a river flowed westwards, but at a far later time, when the basalt-eruptions had ceased and the volcanic plain had been already deeply trenched by erosion, yet before the subterranean fires were finally quenched, as the pitchstone of the Scuir abundantly proves. When one reflects upon the enormous denudation of this region, to which more special reference will be made in the sequel, one is not surprised that many connecting links should have been effaced. The astonishment rather arises that so continuous a story can still be deciphered. Even, however, had the original record been left complete, it would have been exceedingly difficult to trace the successive mutations of a river-channel during long ages of volcanic eruptions. Such a channel would have been concealed from view by each lava-stream that poured into it, and would not have been again exposed save by the very process of erosion that destroys while it reveals. While, therefore, there is not and can never be any positive proof that in the fluviatile records of Canna, Sanday and Eigg successive phases are registered in the history of one single stream, I believe that this identity is highly probable. It was a river which seems to have risen among the mountains of Western Inverness-shire, and it had doubtless already taken its course to the sea before any volcanic eruptions began. It continued to flow westwards across the lava-floor that gradually spread over the plains. Its channel was constantly being filled up by fresh streams of basalt or deflected by the uprise of new cinder-cones. But, fed by the Atlantic rains, it maintained its seaward flow until the general subsidence which carried so much of the volcanic plain below the sea. Yet the higher part of this ancient water-course is no doubt unsubmerged, still traversing the schists of the Western Highlands as it has done since older Tertiary time. It may, perhaps, be recognized in one of the glens which carry seaward the drainage of the districts of Morar, Arisaig, or Moidart. [Illustration: Fig. 280.--View of the Scuir of Eigg from the South.]