declines in that direction from 80 to 40 inches. Thus, in the Glacial Epoch, the snowfall on the former district would be nearly double that on the latter, and great glaciers might be produced in the one, while they would be small or even wanting in the other. This no doubt is true to a certain extent, but the dissentients submit that the argument does not really overcome the difficulty. The change from a precipitation amounting to 80 or 100 inches on the western district to one of half the amount on the eastern, is not sudden, but gradual, even if it be somewhat rapid, so that the rainfall on the hills above Glen Roy can hardly be less than 60 inches, and may be rather greater. Moreover, the difference in elevation between the two districts

is not very great. Though Ben Nevis rises to a height of 4406 feet, it considerably overtops its neighbours, while much of the upland around Glen Roy lies rather above the 2000-foot contour line. Thus, the average height of the two districts can hardly differ by so much as 1000 feet, which corresponds with a disparity in mean temperature amounting to not more than 4° F. Hence, even if allowance be made for the heavier snowfall, the glaciers of the western region would not be large when the head of Glen Roy was free from ice; and if the former became enormous, then Glen Roy itself would be occupied by ice-streams from the surrounding uplands, and its glaciers might even descend to the valley of the Spean.

Professor Prestwich's hypothesis1 avoids this difficulty. He supposes that, at the end of the time of most severe cold,2 "the old ice-sheet of Scotland became covered with pools and lakes, which would go on filling until the water reached the lip of the basin, when the surplus waters would escape along the natural lines of drainage, either on or beneath the ice, to some lower levels; but when, owing to exceptional circumstances, such as those which prevailed in the Spean and Roy valleys, the ice had been heaped up in larger masses so as to raise high the water-level, then these lakes, dammed back pending the removal of the barriers, made their temporary outflow over any 'cols' lying at a lower level than the barriers, and leading into adjacent valleys free from such blockages. So while every mountain-side was contributing its rills and rivulets to innumerable temporary pools and tarns on the melting ice, their waters were, in such instances as those presented by the Lochaber valleys, retained and formed lakes ultimately extending the length and depth of the glens (p. 695).

" 3

1 Phil. Trans. Roy. Soc., 1879, Part II. p. 663.

2 He considers the roads to have been formed at the end of the first or coldest part of the Glacial Epoch, while Mr. Jamieson assigns them to a rather late stage in that epoch.

Professor Prestwich thinks that the "roads" were formed, not by the erosive action of the waves of these lakes, but by the slipping of the loose materials on the hillsides, owing to the sudden removal of support when one of the barriers gave way and the waters escaped. This is almost a necessary part of the hypothesis, for in such lakes the erosive action of waves would be very slight.

Lakes of this kind undoubtedly exist. There are two in the upper part of the Gorner Glacier, just at the foot of Monte Rosa; but even the larger of them is not more than about a couple of hundred yards across in any direction. Professor Prestwich quotes some instances from the great glaciers in the Himalayas; but even these are comparatively small, the largest mentioned being not a third of a mile long; while, though both the Himalayas and Greenland are better known than when his paper was written, no lake of this kind has been discovered which is in any way comparable in size with that which the hypothesis demands. It is also difficult to see how a glacier, under the circumstances alleged, could be sufficiently solid to retain so large a sheet of water as would be required to form these roads. Passing over other difficulties of a less general character, it may suffice to say that the hypothesis, notwithstanding its ingenuity, and the fact that it avoids some serious objections to which the other mode of attributing a fresh-water origin to the roads is exposed, does not appear to have met with general favour.

To conclude the difficulty of the almost unique. occurrence of the roads is common in reality to all the explanations; each has to postulate something exceptional in the conditions of the district. The principal difficulties in the way of the marine origin are two in number. One, that the roads require a submergence during the Glacial Epoch amounting to

more than 1150 feet. The possibility of this opens a wide question, not only in regard to terrestrial physics, but also as to the interpretation of the glacial deposits in the British Isles. This question will be noticed in a later chapter. If it can be proved that a submergence to more than the above amount actually took place on the southern side of the Border, there would be nothing startling in its effects extending to Western Scotland. If, however, the deposits in question testify to elevation rather than to depression, then the objections to the marine hypothesis become almost insuperable. The other difficulty, that of the absence of any marine organism, both in the roads and at any other like elevation in Scotland, though it may be explained, as has been mentioned above, is undoubtedly a grave one. The main objections to the fresh-water hypothesis (apart from the question of a submergence) turn upon the difficulty of accounting for the existence of a mass of ice huge enough to act as a dam for such a body of water, under circumstances which would allow of Glen Roy, with parts of some neighbouring valleys, being free from ice.1 Hence, as there are difficulties on all sides, the history of the Parallel Roads of Glen Roy is likely to remain for the present among the controversial questions of geology.

Much has been written about certain ridges of

1 The gravity of this will be better appreciated after a study of some of the later chapters of this book.