An uncertainty however arises in estimating \(S_{1}\) and \(S_{2}\) for the rate of leak in \(A\) and \(B\) is made up of the current due to emanation alone and the current produced in the gas by the excited radioactivity on the electrodes. As the amount of excited radioactivity increases with the time, the ratio of the current due to the emanation and to the excited radiation varies with the time allowed for diffusion. The ratio of the current due to the excited radiation can be determined by removing the central electrode and finding the amount of current immediately after the introduction of a new electrode. When the emanation is allowed to diffuse for half an hour, the current due to excited radioactivity was about \(\cdot 4\) of the whole. The calculated value of \(K\) was found to be about 20 per cent greater when the correction for the amount of excited radioactivity was applied. The value of \(K\) deduced from the experiments was found to be between \(\cdot 08\) and \(\cdot 15\). All the later observations gave a value about \(\cdot 08\). This variation in the value of \(K\) deduced from the experiments is not altogether due to errors of experiment, the values obtained at first with a new specimen of radium were in all cases higher than when it had been laid by for several months. It appears as if the emanation were not simple in character, and that part of the emanation first given off was of lower molecular weight than that emitted after several months exposure to the air. Further experiments are now being carried out to see if the radium emanation undergoes a progressive change with time. For the purpose of comparison, we will now give a few of the coefficients of inter-diffusion of gases, compiled from Landölt and Bernstein's tables. \begin{array}{|l|l|l|} \hline Gas or vapour & Coefficient of diffusion into air. & Molecular weight \\ \hline Water vapour & \(0 \cdot 198\) & 18 \\ \hline Carbonic acid gas & \(0 \cdot 142\) & 44 \\ \hline Alcohol & \(0 \cdot 101\) & 46 \\ \hline Ether & \(0 \cdot 077\) & 74 \\ \hline \end{array} In the above table we see that the coefficient of inter-diffusion follows the inverse order of the molecular weights. In cases of the simpler gases it has been shown experimentally that the coefficient of inter-diffusion is approximately inversely proportional to the square root of the product of the molecular weight. If we apply these considerations to the emanations \((K= \cdot 08\, \text{to}\, \cdot 15)\) we see that it is a gas or a vapour of molecular weight (allowing a wide margin) probably lying between 40 and 100. These numbers exclude the possibility of the substance being a vapour of radium, for it has already been shown by M. and Mme. Curie that the atomic weight of radium is greater than that of barium. We must therefore conclude that the emanation is in reality a heavy radioactive vapour or gas.