the freezing-point, divided by 1.85, that should be expected if no dissociation took place. As the figure shows, all the curves for the non-conductors—in this case cane-sugar, propyl-alcohol and phenol—converge towards unity with diminishing concentration. At higher concentrations there occur deviations from the simple law. As examples of binary electrolytes are Fig. 6. chosen LiOH, NaCl and LiCl—their curves all converge towards the number 2. As ternary electrolytes are chosen K2SO4, Na2SO4, MgCl2 and SrCl2, they are decomposed into three ions, and their curves therefore all converge towards the number 3.
As I had taken a step that seemed most adventurous to chemists, there remained to investigate its chemical and physical consequences. The most general and wide-reaching of these is that the properties of a highly attenuated solution of an electrolyte ought to be additive, that is, composed of the properties of the different ions into which the electrolyte is decomposed. This was already known to be the case in many instances, and Valson had to this end tabulated his 'modules' by the addition of the one value for the negative to the other for the positive ion, we may calculate the properties of any electrolyte composed of the tabulated ions. In this way we may treat the specific weight (Valson), the molecular conductivity (law of Kohlrausch), the internal friction (Arrhenius), the capillarity (Valson), the compressibility (Röntgen and Schneider), the refractive index (Gladstone), the natural rotation of polarization (law of Oudemans), the magnetic rotation of polarization (Perkin and Jahn), the magnetization (Wiedemann), and all other properties of the electrolytes hitherto sufficiently studied.
The most important of these additive properties are those of which we make use in chemical analysis. As is well known, it is generally true that chlorides give a white precipitate with silver salts. It was said formerly that silver salts are reagents for chlorine. Now we say that silver ions are reagents for chlorine ions. This expression is better than the old one, for neither all silver salts, e. g., potassium silver cyanide and many other compounds of silver, nor all chlorine compounds, e. g., potassium chlorate and many organic chlorides, give this characteristic reaction. The experiment succeeds only with such