The sides of all canyons in an arid region are more or less protected in the same way.  That is, the rains fall suddenly, rarely continuously for any length of time, and are collected and conducted away immediately, not having a chance to enter the ground.  Homogeneous sandstone preserves its perpendicularity better than other rocks, one reason being that it does not invite percolation, and usually offers, for a considerable distance on each side of the canyon, barren and impervious surfaces to the rains.  Where strata rest on exposed softer beds, these are undermined from the front, and in this way recession is brought about.

* Just as wheat flour getting wet on the surface protects the portion below from dampness.  The rainfall is often so slight, also, that a surface is unchanged for years.  I once saw some wagon tracks that were made by our party three years before.  From peculiar circumstances I was able to identify them.

** Robert Brewster Stanton explained this very clearly in his investigations for the Canadian Pacific Railway into the causes of land-slides on that line.

In the basin of the Colorado are found in perfection all the extraordinary conditions that are needed to bring forth mammoth canyons.  The headwaters of all the important tributaries are invariably in the highest regions and at a long distance from their mouths, so that the flood waters have many miles of opportunity to run a race with the comparatively feeble erosive forces of desert lands.  The main stream-courses are thus in the lower arid regions and in sedimentary formations, while their water-supply comes from far away.  The deepest gorges, therefore, will be found where the rainfall is least, unless diminishing altitude interferes.  Thus the greatest gorge of the whole basin, the Grand Canyon, is the one farthest from the sources of supply, and in the driest area, but one, of the whole drainage system.  It ends abruptly with the termination of the high arid plateau which made it possible, but had this plateau extended farther, the Grand Canyon would also have extended a similar distance.  It is plain then that the cutting of these canyons depends on the amount of water (snow is included) which may fall in the high mountains, the canyons themselves being in the drier districts.  It is also clear that if, by some chance, the precipitation of the high sources should increase, the corrasion of the stream-beds in the canyons would likewise increase and outrun with still greater ease the erosion of their immediate surroundings.  On the other hand, if the precipitation in the arid surroundings should increase, the wearing down of the side walls would for a time—­till covered by debris and vegetation—­go on more rapidly till, instead of Canyons of the Colorado River type, there would be deep, sharp valleys, or wide valleys, according to the amount of difference between the precipitation of the low lands and the