(3) That as the wires A and B are clamped separately below, we may impart a sudden molecular disturbance to either A or B by giving a quick to-and-fro (torsional) vibration round the vertical wire, as axis, by means of the handle. As the wire A is separate from B, disturbance of one will not affect the other. Vibration of A produces a current in one direction, vibration of B in the opposite direction. Thus we have means of verifying every experiment by obtaining corroborative and reversed effects. When the two wires have been brought to exactly the same molecular condition by the processes of annealing or stretching, the effects obtained on subjecting A or B to any given stimulus are always equal (fig. 56). [Illustration: FIG. 56.--EQUAL AND OPPOSITE RESPONSES EXHIBITED BY A AND B] Usually I interpose an external resistance varying from one to five megohms according to the sensitiveness of the wire. The resistance of the electrolyte in the cell is thus relatively small, and the galvanometer deflections are proportional to the E.M. variations. It is always advisable to have a high external resistance, as by this means one is not only able to keep the deflections within the scale, but one is not troubled by slight accidental disturbances. #Graduation of intensity of stimulus.#--If now a rapid torsional vibration be given to A or B, an E.M. variation will be induced. If the amplitude of vibration be kept constant, successive responses--in substances which, like tin, show no fatigue--will be found to be absolutely identical. But as 'the amplitude of vibration' is increased, response will also become enhanced (see Chap. XV). [Illustration: FIG. 57.--TOP VIEW OF THE VIBRATION CELL The amplitude of vibration is determined by means of movable stops S S', fixed to the edge of the graduated circle G. The index arm I plays between the stops. (The second index arm, connected with B, and the second circle are not shown.)] Amplitude of vibration is measured by means of the graduated circle (fig. 57). A projecting index, in connection with the vibration-head, plays between fixed and sliding stops (S and S'), one at the zero point of the scale, and the other movable. The amplitude of a given vibration can thus be predetermined by the adjustment of the sliding stop. In this way we can obtain either uniform or definitely graduated stimuli.