When the level of a rivulet allows, and the waters are simply surface discharge, the system has been used of making two open channels, one upon each side of a steep valley, thus retaining the waters, allowing the adoption of two short span open culverts and two channels instead of one large culvert at the deepest place, and saving expense, the original bed of watercourse being filled with the surplus excavation. The embankment consisting of broken rock or hard granular soil, any little percolation of water along the old bed will not deleteriously affect it, but will find a passage. CHAPTER V. APPROXIMATE SAFE MAXIMUM LOAD UPON DIFFERENT EARTHS.—NORMAL PRESSURE OF THE EARTH.—THE SAFE MAXIMUM LOAD UPON DEPOSITED EARTH.—APPROXIMATE SAFE MAXIMUM HEIGHT OF AN EMBANKMENT. There is no limit to the depth of a cutting except a due regard to economical construction, provided the slopes are sufficiently flat, and the lateral and upward fluid pressure in the slope, and formation and quantity of water not too great for the stability of the earth; but, in the case of embankments, the load upon the ground and the deposited material in great measure restricts the height, or necessitates an embankment being gradually spread out, so as to enlarge the bearing area as the weight is increased. The following values of the safe maximum compressive load have been compiled from actual practice, but are, of course, only intended as a guide to the safe load for foundations excavated in ground not artificially deposited. The condition of the earth in each case should be considered, and in works of magnitude it is advisable to make experiments extending as long as practicable, and for at least a month, and it is false economy not to carefully ascertain the character, condition and other circumstances of a foundation destined to support any part of a structure, a failure of which may result in serious consequences; and it should be borne in mind that a continuous surface possesses greater sustaining power than the same area in detached portions, as the adhesion of the sides is not destroyed; similarly, the load that a tenacious earth will support upon a small area is somewhat greater than over a large area, because the lateral surfaces are relatively larger in proportion to the area, and, therefore, the effect of cohesion is proportionately greater; but in loose soils it is not so, for cohesion exists but in name, and the ground around would be upheaved upon an excessive load being superimposed. In testing the weight any earth will support, it is not so much the first settlement, provided it is not excessive, that it is desirous to know, but whether after the first settlement it ceases or the earth, as it were, reacts and rebounds, which it may do in firm ground to the extent of one-eighth to half an inch. If so, the ground is not overloaded.