A completely decisive proof that argon is not a mixture has just been furnished by experiments carried out by Dr. Collie and Professor Ramsay, in which a large quantity of argon was submitted to fractional diffusion. From what was said on p. 162, it will be seen that if argon consisted of a mixture of two gases of different densities, such a process should separate the mixture more or less completely into its two constituents. After a long series of diffusions, however, the density of that portion of argon passing first through the porous plug, which would have been less had any gas of lower density been present, was found to be identical with that of the last portions of gas. On the other hand, by aid of the same diffusion-apparatus, a fair separation of oxygen (density 16) from carbon dioxide (density 22) was effected, although, as the reader will observe, the densities of these two gases do not differ greatly. Hence, if argon consists of two kinds of matter they must have the same density, and hence the same molecular weights, and the difficulty is not removed. But as the spectrum of the first and last portions was the same and was identical with that of argon, this supposition is improbable. The evidence is therefore distinctly against the supposition that argon is a mixture of two or more elements. There is, however, another possible method of accounting for the high atomic weight of argon, which, if it could be reduced by a few units, would fall into its place after chlorine and before potassium. It is that argon consists of a mixture of many monatomic, with comparatively few diatomic, molecules. If there were only about 500 molecules of diatomic argon in every 10,000 molecules of the gas, its density, supposing it to consist entirely of monatomic molecules, would be 19, and its atomic and molecular weights 38, a number which would fit between the atomic weight of chlorine, 35·5, and that of potassium, 39·1. Several instances of this kind are known. Chlorine itself, when heated to high temperatures, changes from diatomic to monatomic molecules, and the density decreases with the change. For example, at 1000° the found density of chlorine is 27, implying a molecular weight of 54; now 54 is neither the weight of a monatomic molecule of chlorine, viz. 35·5, nor of a diatomic molecule, which is 71; but it corresponds to that of a mixture of monatomic and diatomic molecules. Here fall of temperature causes combination of monatomic molecules with each other to form diatomic molecules; and rise of temperature increases the number of monatomic molecules, at the expense of the diatomic molecules. Is there no sign of similar behaviour with argon?