We obtained the same results again when the disc M M is replaced by a Faraday cylinder filled with air, and covered outside with insulating material. The opening of the cylinder, closed by the thin insulating plate, _p p_, was opposite the radiating source. [Illustration: FIG. 7.] Finally, we made the inverse experiment, which was to place the lead receptacle with the radium in the centre of the insulating material and in connection with the electrometer (Fig. 7), the whole being surrounded with the metallic covering connected to earth. Under these conditions, it is evident from the electrometer that the radium has a positive charge equal in magnitude to the negative charge of the former experiment. The radium rays penetrate the thin dielectric plate, _p p_, and leave the conductor inside carrying with them negative electricity. The α-rays of radium do not interfere in these experiments, being almost completely absorbed by a very thin layer of matter. The method just described is not suitable for the study of the charge of the rays of polonium, these rays very slightly penetrating. We observed no indication of any charge in the case of polonium, which gives rise to α-rays only; but, for the reason just given, no conclusion can be drawn from this. Thus, in the case of the deflected β-rays of radium, as in the case of cathode rays, the rays carry a charge of electricity. But, hitherto, the existence of electric charges uncombined with matter has been unknown. In the study of the emission of the β-rays of radium, we are therefore led to make use of the theory which is in vogue for the study of cathode rays. In this ballistic theory, formulated by Sir William Crookes, since developed and completed by Prof. J. J. Thomson, the cathode rays consist of extremely minute particles, which are hurled from the cathode with great velocity, and which are charged with negative electricity. We might similarly conceive that radium sends into space negatively electrified particles. A specimen of radium, enclosed in a solid thin perfectly insulated envelope, should become spontaneously charged to a very high potential. By the ballistic hypothesis the potential would increase until the potential difference of the surrounding conductors became sufficient to hinder the ejection of the electrified particles and to cause their return to the source of radiation. We have performed an experiment on these lines. A specimen of very active radium was enclosed for some time in a glass vessel. In order to open the vessel, we made a trace on the glass with a glass cutter. Whilst so doing, we clearly heard the report of a spark, and upon examining the vessel with a magnifying glass, we observed that the glass had been pierced by a spark at the spot where it had been weakened by the scratch. The phenomenon produced is comparable to the rupture of the glass of an overcharged Leyden jar.