This preliminary operation does not yield pure argon; it merely removes a large portion of the nitrogen. To free the argon from the remainder, it is caused to circulate (by means of a specially contrived mercury-pump, where each drop of mercury in falling down a narrow glass tube carries before it a small bubble of gas) through tubes containing red-hot copper, red-hot copper oxide, red-hot magnesium, and cold soda-lime and phosphoric anhydride. The copper serves to remove traces of oxygen; the copper oxide yields up its oxygen to any hydrogen or carbon compound--dust and the like--which may happen to be present; the soda-lime absorbs any carbon dioxide produced by the combustion of the carbon compounds, and at the same time partially dries the gas; while the phosphoric anhydride effectually dries the gas, previous to its passage over the red-hot magnesium, which in its turn removes the nitrogen. It is necessary to continue this circulation for several days before the litre of gas is entirely freed from nitrogen. It is difficult to choose between these two methods: both are troublesome, and require a considerable time, but in an ordinary laboratory the latter is probably the more easily set in operation, for the former requires a suitable electric current, and power, so as to rotate the water-fan. Up to the present date, the only sources which have yielded argon are atmospheric air, gases extracted from mineral waters or from springs, one meteorite, and a few rare minerals. No animal or vegetable substance appears to contain it. Experiments were made in the summer of 1895 by Mr. George MacDonald and Mr. Alexander Kellas, in order to decide whether argon was a constituent of any living matter. Some peas were reduced to powder and dried; the carbon and hydrogen of the peas were burned to carbon dioxide and water by heating with oxide of copper, and under these circumstances the nitrogen is evolved in the state of gas. Had argon been contained in the vegetable, it too would have accompanied the nitrogen. The nitrogen was then, as usual, absorbed for the most part by means of magnesium, and the small unabsorbed residue was mixed with oxygen and exposed to electric sparks for many hours, in presence of caustic soda. There was _no_ residue left after absorbing the excess oxygen: the gas was completely removed. Similar experiments carried out on animal tissue led to a similar conclusion. Two mice were chloroformed, and when dead they were dried in an oven until all the moisture of their bodies was completely driven off, and it was possible to reduce them to powder. It is interesting to note that one of these mice contained 73 per cent of water, and the other 70·5 per cent. The dried animals yielded about 11 per cent of their weight of nitrogen. Absolutely no residue of gas was obtained on causing this nitrogen to combine; hence it appears to be a legitimate conclusion that neither animal nor vegetable tissue contains any appreciable amount of argon.