Page:Transactions of the Geological Society, 1st series, vol. 4.djvu/389

 on the borders of Loch Lochy, as well as on the principal extent of the borders of Loch Laggan, and the valley of the Spean. There is a set of common features through the whole tract, the same rocks, the same slopes, the same causes of waste, yet the water marks are strongly defined through a portion of this wide space, while they are totally wanting in others. If we increase the extent of the supposed lake, we increase this difficulty, but it is already sufficiently great. It is proper to notice that this objection to the hypothesis of a lake, is equally applicable to the other two modes offered for explaining the action of water in producing these lines. If it is held sufficient to reject the one, it is equally valid against the other two, although it seemed superfluous to enumerate it among the more serious difficulties to which those are liable.

The complete and sudden transition from the uppermost line of Glen Roy to the next succeeding one, and finally to the present bottom of the valley or perfect draining of the whole, shows that the lake which occupied these vallies had subsided at three different intervals. As it is a more probable supposition that these three successive trainings took place at the same point, than at different ones, it will be most convenient to assume the present and lowest communication, namely, the real exit of the waters of the Spean and Roy, as that point. This place must be situated on the river Lochy, before its junction with the sea at Loch Eil. Here then we must imagine that a dam has subsisted, not gradually worn down by the slow corroding action of the river issuing from the lake, but by three successive failures occurring suddenly, or at least within short intervals of time. Had much time elapsed between these intervals, the several lines must have been more obscurely marked, or intermediate ones of smaller dimensions must have been visible. It is plain that the difficulties will be unnecessarily complicated if we consider