The gas was then weighed, this time in a larger bulb, the weight being 0·2190 gram; and such is the possibility of precision in weighing on a good balance, that a difference of one two-thousandth of the whole weight was detectable. The density of the gas was now found to be 16·1. At this stage it was still believed that the new gas was an ozone-like modification of nitrogen, difficult to attack by magnesium. It was supposed that just as oxygen, when exposed to an electric discharge, undergoes a cleavage of its molecules, two-atom molecules becoming one-atom molecules for an instant, which then unite to form three-atom molecules, so the action of the magnesium on the nitrogen might be to withdraw one atom of nitrogen from the two-atom molecule, leaving a single uncombined atom, which might not improbably find two partners, each of its own kind, to form with them a three-atom molecule--a sort of nitrogen-ozone, in fact. Hence it was resolved to continue the absorption with fresh magnesium for a still longer time, in the hope of its being possible to isolate the three-atom nitrogen molecules. But it became apparent that the bright metallic magnesium was now not much attacked; and on estimating the total amount of nitrogen absorbed, by treating the compound of nitrogen and magnesium with water, and liberating the nitrogen as ammonia, it appeared that only a small quantity of magnesium nitride had been formed. The density of this further purified gas was again determined, when it was found that a litre now weighed 1·7054 gram, corresponding to a density of 19·086. A portion of this gas was mixed with oxygen and exposed to a rain of electric sparks in presence of caustic soda; in fact, Cavendish’s old plan of causing nitrogen to combine was now resorted to. Contraction occurred, and on removing the excess of oxygen, the diminution of volume was found to amount to 15·4 per cent of the original volume taken. Making the supposition that the gas of density 19 still contained nitrogen, and allowing for its influencing the density, it followed that the pure gas should be twenty times as heavy as hydrogen.