The paper or card on which the diagram is drawn is wrapped around a brass cylinder, _A A_. This cylinder is made to revolve part of the way around by a strong twine, _a b_, which is wrapped around a pulley, _b_, at the bottom of the cylinder. The twine is attached to a lever, similar to that shown in fig. 30, which receives a reciprocating motion from the piston of the engine. The twine can of course move the cylinder only in one direction, and therefore a coiled spring similar to a clock spring is placed inside of the cylinder to draw it back when the twine is relaxed. In this way the paper cylinder or drum receives a part of a revolution at each stroke of the piston, and moves simultaneously with it. This drum is used instead of a flat card, on account of the practical difficulties of employing the latter. The motion of the paper on this drum will, however, be exactly the same in relation to the pencil as the motion of a flat card would be. [Illustration: Fig. 136a.] The method of attaching an indicator to a locomotive is represented in fig. 136a. It will be seen from this that it is placed over the center of the steam chest and connected to each end of the cylinder with ³⁄₄-inch pipes. A globe valve was in the case represented placed on each side of the indicator, so that it could be put into communication with either end of the cylinder, or could be completely shut off from both. A better plan, however, is to have a three-way cock at the point where the horizontal pipe connects with the vertical one leading to the indicator, as the passages in a three-way cock are more direct than those in globe valves. The arrangement of the levers for giving motion to the indicator drum, and of the seat, which is very requisite for the experimenter, will be readily understood from the engraving without further explanation. It is thought by some engineers that the indicator should be applied as near to each end of the cylinder as possible. It is believed, however, that if the pipes, cocks, and their connections are made large enough so as not to impede the motion of the steam, no appreciable error will arise from the method illustrated in fig. 136a. QUESTION 216. _What should be the form of an indicator diagram, if the steam is distributed by a link motion so as to produce the best practicable action in the cylinders?_