The gases can be almost entirely extracted by boiling the water. But to boil large quantities of water at one operation in a vessel suitable for collecting the escaping gas is not easy. It is much simpler to cause the water to pass slowly through a can below which there is a powerful flame, so that the water in its passage becomes heated to the boiling-point, and gives off its gas before it escapes. Of course the gas collected contained oxygen, but this was easily removed by the usual method of passing it over red-hot copper. The density of the residual gas was determined, and it was found to be at least as much greater than that of “atmospheric” nitrogen as the density of “atmospheric” nitrogen exceeded that from chemical sources. Hence it was to be concluded that the new constituent of air, argon, was being concentrated by dissolving air in water, and extracting the dissolved mixture of gases. A third proof that argon exists in air will be given farther on. [Illustration: FIG. 3.] In order that the properties of the newly-discovered gas, argon, might be thoroughly investigated, it was necessary to prepare it on a much larger scale than had hitherto been attempted, and this was carried out by the two processes for removing the oxygen and nitrogen which have been already described. Supposing the new gas to have the density 20 compared with oxygen as 16, the density of the atmospheric mixture of nitrogen and argon compared with that of nitrogen alone shows that air should, roughly speaking, contain less than one part of argon in one hundred. Hence, to obtain a litre of argon, it was necessary to work up a large quantity of atmospheric nitrogen. Now, as has just been said, there are two ways of doing this. (1) One is to produce an electric flame between two pieces of stout platinum in air, confined in a large glass balloon of about 6 litres capacity, over a weak solution of caustic soda. For this purpose a very powerful rapidly alternating current is necessary. The latest, and apparently the best, method of carrying this out, was described by Lord Rayleigh in his Royal Institution lecture in January 1896. The neck of the balloon is placed downwards, and connected by means of a glass tube, passing through a cork which closes the neck, with a rotating fan or paddle-wheel with curved blades, which forces through the tube a weak solution of caustic soda; another tube, also entering through the cork, conveys away the excess of soda to the fan, whence it is again forced into the balloon. The soda solution makes a fountain in the balloon, and flows in a uniform stream down its sides, covering its inner surface with a thin layer of liquid. Through the cork the two electrodes, with their thick platinum terminals, enter; and there is another tube besides, which conveys into the balloon a mixture of air and oxygen in such proportions that they combine completely on exposure to the flame. The layer of soda solution plays a double part. It prevents the