With a pressure of three tons per square foot, the 4-in. planks were forced down into the embankment a little more than 6 ins., resulting in a very slight bulging of the slopes a little below the water level. Immediately under the planks the soil became hard and compact. A man’s weight pushed a sharp steel rod, ¾-in. in diameter, only 6 to 8 ins. into the embankment where the pressure was applied, while outside of this area the rod was easily pushed to the bottom of the tank. These results corroborate in a general way the practical experience of the author, both in compressed embankments, where he found it necessary to use a pick vigorously to loosen the material of which they were composed, and in embankments made by merely dumping the material from a track, in which case the earth is so slightly compressed that an excavation is easily made with a shovel. [Illustration: Fig. 21.] [Illustration: Fig. 22.] [Illustration: Fig. 23.–CAN FOR DETERMINING FRICTIONAL RESISTANCE] [Illustration: Fig. 24.] [Illustration: Fig. 25. FIGS. 21 TO 24.–EXPERIMENTAL DIKES AND CYLINDER EMPLOYED IN STUDIES FOR THE NORTH DIKE OF THE WACHUSETT RESERVOIR; AND (FIG. 25) CROSS-SECTION OF THE DIKE.] The difference in the coefficient of friction of the same material when dry and when wet greatly modifies the form of slope. The harder and looser the particles, the _straighter_ will be the slope line in excavation and slips. The greater the cohesion of the earth, the _more curved_ will be the slope, assuming a parabolic curve near the top–the true form of equilibrium. RATE OF FILTRATION.–The rate of filtration through different soils was experimented with by forming a dike in the tank previously mentioned, as shown in Fig. 22. The dike was made full 8 ft. high, 7 ft. wide on top, with a slope on the up-stream side of 2 on 1, and on the down-stream side 4 on 1. This gave a base width of 55 ft. Immediately over the top of the dike there was placed 3 ft. of soil to slightly consolidate the top of the bank and permit the filling of the tank to the top without overflowing the dike. The water pressure in different parts of the dike was determined by placing horizontal pipes through the soil crosswise of the tank. These pipes were perforated and covered with wire gauze, being connected to vertical glass tubes at their ends. The end of the slope on the down-stream side terminated in a box having perforated sides and filled with gravel, thus enabling the water to percolate and filter out of the bank without carrying the soil with it.