water-foods, for power and for navigation. Here is a good work—soil-production—followed by advantageous courses of the water both up and down. On the other hand, the rainfall may rush away on the surface as a foul erosive flood, wasting soil and plant-food, gullying the surface, flooding the valleys, filling the reservoirs, sweeping out the dams, barring the streams and clogging the deltas. If it shall be found that nearly all the rainfall should go into the soil and thence into the under-drainage, coming out slowly and steadily by seepage and by springs into the streams, clear and pure, these streams should present nearly ideal conditions for water-food, for water-power and for stream-navigation. An ideal solution of the soil problem may therefore solve the greater part of the whole complex of problems of which navigation is the last term. It may thus prove to be the key problem. It is clearly the initial problem, for it attacks the rainfall when it first touches the earth.
To see more definitely if it be the key problem, we must turn to details, and yet, with the brevity that is imperative, we may only look at major details, passing by a multitude of special cases, some of which are even exceptions.
While soils are formed by the atmosphere and its waters acting upon rock (aided by plants and animals), soil surfaces are carried away by wind and wash. At any instant, then, the depth of the soil measures the lag of removal behind production. We hasten to note that the addition of new soil below and the loss of exhausted soil above are alike tributary to permanent fertility, and clearly the best results spring from the proper ratio of addition at the bottom, to wastage at the surface.
We have as yet no accurate measure of the rate of soil production. We merely know that it is very slow. It varies obviously with the kind of rock. Some of our soils are derived from material already reduced to a finely pulverized condition. Such are the lowland accumulations from highland wash. Such also is the glacial drift, rock-flour rasped from the face of the rock by the glacial file and ground up with old soils. Soils may be developed from such half-prepared material with relative rapidity, but observation shows that even in these cases, when the slope is considerable, wind, wash and cropping remove the surface much too fast for stable fertility. For average rock, under the usual conditions of our climate, the common estimate of natural loss and gain has been a foot in 4,000 to 6,000 years, which includes channel-cutting and bank-undermining. This seems to me too rapid a rate for ordinary soil production under normal conditions. Without any pretensions to a close estimate, I should be unwilling to name a mean rate of soil-formation greater than one foot in 10,000 years, on the basis of observation. If we allow 40,000 years for the