In order to convey a sufficient quantity of emanation into the half-cylinder _A_, it was necessary to heat the radium slightly. The slide _S_ was closed and the side tubes opened. A slow current of dry air from a gasometer was passed through a platinum tube, in which a small quantity of radium compound was placed. The emanation was carried with the air into the cylinder _A_. When a sufficient quantity had been introduced, the stream of air was stopped. The side tubes were closed by fine capillary tubes. These prevented any appreciable loss of gas due to the diffusion, but served to keep the pressure of the gas inside _A_ at the pressure of the outside air. The three entrance tubes into the cylinder, shown in the figure, were for the purpose of initially mixing the emanation and gas as uniformly as possible. After standing several hours to make temperature conditions steady, the slide was opened, and the emanation began to diffuse into the tube _B_. The current through the tubes _A_ and _B_ was measured at regular intervals by an electrometer, with a suitable capacity in parallel. Initially there is no current in _B_, but after the opening of the slide, the amount in _A_ decreased and the amount in _B_ steadily increased. After several hours the amount in each half is nearly the same, showing that the emanation is nearly uniformly diffused throughout the cylinder. It can readily be shown[255] that if _K_ = coefficient of diffusion of the emanation into air, _t_ = duration of diffusion experiments in secs., _a_ = total length of cylinder, _S₁_ = partial pressure of emanation in tube _A_ at end of diffusion, _S₂_ = partial pressure of emanation in tube _B_ at end of diffusion, then $$ \frac {S_1 − S_2} {S_1 + S_2} = \frac {8} {\pi^2} (e^{\frac {\pi^2 Kt} {a^2}} + \frac {1}{9}e^{-\frac {9 \pi^2 Kt} {a^2}} + ...) $$ . Now the values of _S₁_ and _S₂_ are proportional to the saturation ionization currents due to the emanations in the two halves of the cylinder. From this equation _K_ can be determined, if the relative values of _S₁_ and _S₂_ are observed after diffusion has been in progress for a definite interval _t_. The determination of _S₁_ and _S₂_ is complicated by the excited activity produced on the walls of the vessel. The ionization due to this must be subtracted from the total ionization observed in each half of the cylinder, for the excited activity is produced from the material composing the emanation, and is removed to the electrodes in an electric field. The ratio of the current due to excited activity to the current due to the emanation depends on the time of exposure to the emanation, and is only proportional to it for exposures of several hours.