From the piston-rod of the indicator a pencil projects horizontally, and its point presses against a sheet of paper wound on a drum, which moves about a vertical axis. This drum is made to move backwards and forwards through a part of a revolution, so that its motion may exactly correspond with that of the piston in the cylinder of the steam engine. Thus, if the piston of the indicator were to remain stationary, a level line would be traced on the paper by the movement of the drum; and if the latter did not move, but the steam were admitted to the indicator, the pencil would mark an upright straight line on the paper. The actual result is that a figure bounded by curved lines is traced on the paper, and the curve accurately represents the pressure of the steam at every point of the piston’s motion. The position of the point of the pencil which corresponds with each pound of pressure per square inch is found by trial by the maker of the instrument, who attaches a scale to show what pressures of steam are indicated. If the pressure per square inch is known, it is plain that by multiplying that pressure by the number of square inches in the area of the piston of the engine, the total pressure on the piston can be found. The pressure does not rise instantly when the steam is first admitted, nor does it fall quite abruptly when the steam is cut off and communication opened with the condenser. When the steam is worked expansively, the pressure falls gradually from the time the steam is shut off. Now, the amount of work done by any force is reckoned by the pressure it exerts multiplied into the space through which that pressure is exerted. Therefore the work done by the steam is known by multiplying the pressure in pounds on the whole surface of the piston into the length in feet of the piston’s motion through which that pressure is exerted. The trace of the pencil on the paper—_i.e._, the _indicator diagram_—shows the pressures, and also the length of the piston’s path through which each pressure is exerted, and therefore it is not difficult to calculate the actual work which is done by the steam at every stroke of the engine. If this be multiplied by the number of strokes per minute, and the product divided by 33,000, we obtain what is termed the _indicated horse-power_ of the engine. The work done per minute is divided by 33,000, because that number is taken to represent the work that a horse can do in a minute: that is, the average work done in one minute by a horse would be equal to the raising of the weight of 1,000 lbs. thirty-three feet high, or the raising of thirty-three pounds 1,000 feet high. The number, 33,000, as expressing the work that could be done by a horse in one minute, was fixed on by Watt, but more recent experiments have shown that he over-estimated the power of horses, and that we should have to reduce this number by about one-third if we desire to express the actual average working power of a horse. But the