The first two determinations were made with chlorides, of which one was 230 times and the other 600 times as active as uranium. These two experiments gave the same figure as the experiment with the pure barium chloride. There was therefore no hope of finding a difference except by using a much more active product. The following experiment was made with a chloride, the activity of which was about 3500 times as great as that of uranium; and this experiment enabled me, for the first time, to observe a small but distinct difference; I found, as the mean atomic weight of the metal contained in this chloride, the number 140, which showed that the atomic weight of radium must be higher than that of barium. By using more and more active products, and obtaining spectra of radium of increasing intensity, I found that the figures obtained rose in proportion, as is seen in the following table (p. 28). The figures of column A must only be looked upon as a rough estimate. The calculation of the activity of strongly radio-active bodies is difficult, for many reasons which will be discussed later. A. │ M. 3500 │140 Spectrum of radium faint. 4700 │141 7500 │145·8 Spectrum of radium strong, but that of │ barium predominating. ────────────────────┼────────────────────────────────────────────────── Order of Magnitude, │173·8 The two spectra of almost equal intensity. 10^6 ... │ 〃 │225 Only a trace of barium present. A represents the activity of the chloride, that of uranium being unity; M the atomic weight found. At the termination of the processes described above, I obtained, in March, 1902, 0·12 grm. of radium chloride, of which Demarçay made the spectral analysis. This radium chloride, in the opinion of Demarçay, was fairly pure; its spectrum, however, showed the three principal barium lines with considerable intensity. I made four successive estimations of the chloride, the results of which as follows:— Anhydrous radium chloride. Silver chloride. M. I. 0·1150 0·1130 220·7 II. 0·1140 0·1119 223·0 III. 0·11135 0·1086 222·8 IV. 0·10925 0·10645 223·1 I then re-purified this chloride, and obtained a much purer substance, in the spectrum of which the two strongest barium lines were very faint. Given the sensitiveness of the spectrum reaction of barium, Demarçay estimated that the purified chloride contained only the merest traces of barium, incapable of influencing the atomic weight to an appreciable extent. I made three determinations with this perfectly pure radium chloride. The results were as follows:—