It was not inconceivable that if such a “silent electric discharge” were to be passed through “atmospheric” nitrogen, it might increase the number of such three-atom molecules, and might render the gas still denser; or if passed through “chemical” nitrogen, it might increase its density so as to make it equal to that of “atmospheric” nitrogen. Lord Rayleigh made such experiments, but without changing the density in the least: the nitrogen from ammonia or from oxides of nitrogen, which has been termed “chemical” nitrogen, still remained too light by about one two-hundredth, and the atmospheric nitrogen still remained too heavy by the same amount. At this stage Professor Ramsay asked and received permission to make some experiments on the nitrogen of the atmosphere, with the view of explaining its anomalous behaviour. He had several years before made experiments on the possibility of causing nitrogen and hydrogen to combine directly, by passing the mixture over heated metals; among these was magnesium, and although no direct combination to any great extent was observed, still it was noticed that magnesium was a good absorbent for nitrogen, when that gas was passed over the red-hot filings of the metal. This process was therefore applied to the absorption of “atmospheric” nitrogen, in order to find out whether any portion of it was different from the rest. The plan adopted was to heat turnings of magnesium, which can be made very thin and loose, to redness in a tube of hard glass, in contact with the nitrogen of the atmosphere, carefully purified from oxygen, which would otherwise have also combined with the metallic magnesium. As absorption proceeded, more nitrogen was admitted from a reservoir, and after a certain quantity had been absorbed, the residual gas was extracted from the tube by a mercury pump, and weighed. The amount weighed was very small,--smaller perhaps than had up till then been thought possible, if accurate results were to be obtained. But here large differences were to be looked for. Only 40 cubic centimetres--the twenty-fifth part of a litre--was weighed; and its weight was only 0·050 gram. But with careful weighing the error should not exceed one five-hundredth of the amount weighed; and if there were to be any increase in density, that increase should be expected greatly to exceed this small fraction. The first weighing--in May 1894--showed that the nitrogen had increased in density by reason of the operations, and instead of being fourteen times as heavy as hydrogen, it was nearly fifteen times as heavy. [Illustration: FIG. 1.]