My prime mover was a weight supported by a column of sand, which, when the sand was allowed to run out through a variable orifice below, could be made to descend with any desired velocity and yet with uniformity. In addition, by these means an unlimited power could be brought to bear, depending on the size of the weight. Previously it was proposed to use water, and compensate for the decrease in flow, as the column shortened, by a conical vessel; but it was soon perceived that as each drop of water escaped from the funnel-shaped vessel, only a corresponding weight would be brought into play. This is not the case with sand, for in this instance every grain that passes out causes the whole weight that is supported by the column to come into action. In the former instance a movement consisting of a series of periods of rest and periods of motion occurs, because power has to accumulate by floating weight lagging behind the descending water, and then suddenly overtaking it. In the latter case, on the contrary, there is a regular descent, all minor resistances in the slide being overcome by the steady application of the whole mass of the weight. When these advantages in the flow of sand were ascertained, all the other prime movers were abandoned. Mercury-clocks, on the principle of the hydrostatic paradox, air-clocks, &c., in great variety, had been constructed. [Illustration: Fig. 34. The Sand-Clock.] The sand-clock consisted of a tube _a_ (Fig. 34), eighteen inches long and one and a half in diameter, nearly filled with sand that had been raised to a bright red heat and sifted. Upon the top of the sand a leaden weight _b_ was placed. At the bottom of the tube a peculiar stopcock, seen at (2) enlarged, regulated the flow, the amount passing depending on the size of the aperture _d_. This stopcock consisted of two thin plates, fixed at one end and free at the other. The one marked _e_ is the adjusting lever, and its aperture moves past that in the plate _g_. The lever _f_ serves to turn the sand off altogether, without disturbing the size of the other aperture, which, once set to the moon’s rate, varies but slightly in short times. A movable cover _h_, perforated to allow the cord _i_ to pass through, closed the top, while the vessel _k_ retained the escaped sand, which at suitable times was returned into the tube _a_, the weight _b_ being temporarily lifted out. From the clock the cord _i_ communicated motion to the frictionless slide, as shown in Fig. 33. This cord should be as inelastic as possible, consistent with pliability, and well waxed.