He then sought to discover the degree of pressure which it was most expedient to exert between the two adjacent ends of the wire, and for this purpose he effected the pressure by means of weights. He ascertained that when the pressure did not exceed the weight of an ounce on the square inch at the point of connection, the sounds were reproduced with distinctness, but somewhat imperfectly. He next modified the conditions of the experiment, and satisfied himself that it was unnecessary to join the wires end to end in order to obtain this result. They might be placed side by side on a board, or even separated (with a conductor placed crosswise between them), provided that the conductors were of iron, and that they were kept in metallic connection by a slight and constant pressure. The experiment was made with three Paris points, and arranged as it appears in fig. 49, and it has since been repeated under very favourable conditions by Mr. Willoughby Smith with three of the so-called rat-tail files, which made it possible to transmit even the faint sound of the act of respiration.[15] [Illustration: FIG. 49.] He afterwards tried different combinations of the same nature, which offered several solutions of continuity, and a steel chain produced fairly good results, but slight inflections, like those caused by the _timbre_ of the voice, were not reproduced, and he tried other arrangements. He first sought to apply metallic powders to the points of contact; powdered zinc and tin, known in commerce under the name of white bronze, greatly increased the effects obtained; but they were unstable, on account of the oxidation of the contacts; and it was in seeking to solve this difficulty, as well as to discover the most simple means of obtaining a slight and constant pressure on the contacts, that Mr. Hughes was led to the arrangement, previously described, of carbons impregnated with mercury, and he thus obtained the maximum effect.[16] Mr. Hughes considers that the successful effects of the microphone depend on the number and perfection of the contacts, and this is doubtless the reason why some arrangements of the carbon pencil in the instrument described above were more favourable than others.