In a previous theoretical paper Hartley and Robinson have pointed out what very misleading results are obtained in calculating the particle size of a dye from the diffusion coefficient if the Stokes-Einstein equation is used and the electrical forces are neglected. The complicating effect of the electrical forces on the diffusion coefficient has, of course, been realized in researches on other colloidal electrolytes. Thus Svedberg, Tizelius, Northrop, McBain and others have taken them into account in the case of proteins, while McBain and Liu pointed out that the diffusion coefficient of soaps is given by an extension of the Nernst equation. But in experiments with dyes, although many diffusion measurements were reported, this aspect of the matter had been entirely overlooked. An investigation of the diffusion coefficients of dyes is, therefore, of particular interest and also because of the light it throws on the more general problem of determination of the particle size of a colloidal electrolyte, where the particle (unlike that of the protein, which appears to be a simple molecule) consists of a number of ions in association. Many dyes are colloidal electrolytes containing multivalent complex ions. Their diffusion coefficient is consequently largely determined by the mobilities of the ions. It was shown that a minimum value for the diffusion coefficient can be calculated from the conductivity. It follows that all dyes that have high conductivities—this seems to include most, if not all, substantive dyes—must have high diffusion coefficients. Consequently, as was shown, it is not possible to obtain even a qualitative measure of the particle size from the diffusion coefficient and the Stokes-Einstein equation.