(6) The contraction of the earth's interior by cooling and by the emission of material from below the over-crust, has caused this crust to press downward, and therefore laterally, and so to effect great bends, folds, and plications; and these, modified subsequently by surface denudation, and the piling of sediments on portions of the crust, constitute mountain chains and continental plateaus. As Hall long ago pointed out,[26] such lines of folding have been produced more especially where thick sediments had been laid down on the sea-bottom, and where, in consequence, internal expansion of the crust had occurred from heating below. Thus we have here another apparent paradox, namely, that the elevations of the earth's crust occur in the places where the greatest burden of detritus has been laid down upon it, and where, consequently, the crust has been softened and depressed. We must beware, in this connection, of exaggerated notions of the extent of contraction and of crumpling required to form mountains. Bonney has well shown, in lectures delivered at the London Institution, that an amount of contraction, almost inappreciable in comparison with the diameter of the earth, would be sufficient; and that, as the greatest mountain chains are less than 1/600th of the earth's radius in height, they would, on an artificial globe a foot in diameter, be no more important than the slight inequalities that might result from the paper gores overlapping each other at the edges. This thinness of the crushed crust agrees with the deductions of physical science as to the shallowness of the superficial layer of compression in a cooling globe. It is perhaps not more than five miles in thickness. A singular proof of this is seen by the extension of straight cracks filled with volcanic rock in the Laurentian districts of Canada.[27] The beds of gneiss and associated rocks are folded and crumpled in a most complex manner, yet they are crossed by these faults, as a crack in a board may tear a sheet of paper or a thin veneer glued on it. We thus see that the crumpled Laurentian crust was very thin, while the uncrushed sub-crust determined the line of fracture. [26] Hall (American Association Address, 1857, subsequently republished, with additions, as "Contributions to the Geological History of the American Continent"), Mallet, Rogers, Dana, La Conte, etc. [27] As, for instance, the great dyke running nearly in a straight line from near St. Jerome along the Ottawa to Templeton, on the Ottawa, and beyond, a distance of more than a hundred miles.