Page:Whymper - Scrambles amongst the Alps.djvu/91

Rh If a single gradient had ruled throughout, rising from the French to the Italian side, it would have been reduced to 217 in 20,000; but although this would have been of the greatest advantage in working the line, it would have added one more difficulty to the construction of the tunnel. There were enough difficulties without adding another."

"It is, besides, evident that driving the tunnel to a summit doubles the chances of the two ends meeting, and negatives, to a certain extent, the possibilities of error from the two operations upon which the least dependence could be placed,—the triangulation and the levelling. Provided that the two axes are in the same direction, they must meet sooner or later; whether this happens a few yards more to the north or to the south is of no importance."

At the commencement of the tunnel in 1857, there was no accommodation at either end for those employed on the works; and for a long time both engineers and workmen had to submit to numerous privations. Roads had to be made, and barracks to be erected; one after another, houses and shops were added, and at the present time the tunnel-buildings alone form considerable villages at the two ends.

The situations of the two mouths are essentially different from each other. That at Bardonnêche comes out at the bottom of the valley of Rochemolles; that at Fourneaux 300 feet above the Cenis road. At the latter end the debris has been shot out at the mouth down the mountain-side; and, large as the tip (in the language of