The speaker stands before the mouthpiece E, as before a telephone or an acoustic tube, and speaks in a strong, emphatic voice, with his lips pressed against the walls of the mouthpiece, as we see in fig. 67; at the same moment he turns the handle of the cylinder, which is provided with a heavy fly-wheel in order that the movement may be regular. Influenced by the voice, the plate L L begins to vibrate, and sets the tracing point at work, which presses on the tinfoil at each vibration, and produces a furrow whose depth varies along its course in correspondence with the unequal vibrations of the disk. The cylinder which moves at the same time presents the different parts of the groove of which we have spoken to the tracing point in succession; so that when the spoken sentence comes to an end, the design which has been pricked out, consisting of a succession of reliefs and depressions, represents the registration of the sentence itself. The first part of the operation is therefore accomplished, and by detaching the sheet from the instrument the words may be put away in a portfolio. We have now to see how the instrument is able to reproduce what has been so easily inscribed. For this purpose it is only necessary to repeat the process, and the identical effect will be reproduced in an inverse sense. The tracing stylus is replaced at the end of the groove it has already traversed, and the cylinder is again set in motion. When the engraved track passes again under the point, it has a tendency to raise it and to impart to it movements which must necessarily be the repetition of those which first produced the tracing. The vibrating plate is obedient to these movements, and begins to vibrate, thus producing the same sounds, and consequently the same words; yet since there is necessarily a loss of power in this double transformation of mechanical effects, the speaking tube C is attached to the mouthpiece E in order to intensify the effects. Under these conditions the words reproduced by the instrument may be heard in all parts of a hall, and it is startling to hear this voice--somewhat shrill, it must be admitted--which seems to utter its sentences from beyond the grave. If this invention had taken place in the middle ages, it would certainly have been applied to ghostly apparitions, and it would have been invaluable to miracle-mongers. [Illustration: FIG. 68.]