It was next deemed necessary to test whether nitrogen was homogeneous by preparing it too by several different methods. In the same paper Lord Rayleigh (p. 146) mentions that nitrogen, prepared from ammonia, its compound with hydrogen, is somewhat lighter than “atmospheric nitrogen,” the deficiency in weight amounting to about 1 part in 200. Now it is evident from inspection of the numbers quoted above, that the accuracy of the density determination may be trusted to within 1 part in 10,000, and that the balance would detect a discrepancy one-fiftieth of that observed in the densities of “atmospheric” and “chemical” nitrogen. In a letter to _Nature_, Lord Rayleigh asked for suggestions from chemists as to the reason of this curious anomaly, but his letter went without reply. He himself was inclined to believe that the difference was due to the decomposition of some of the ordinary molecules of nitrogen, usually believed to consist of two atoms in union with each other, into molecules consisting of one atom; and as it is held that equal numbers of molecules inhabit the same volume, temperature and pressure being equal, if the total number of molecules in his globe were increased by the splitting of some double-atom molecules into single-atom molecules, the effect would be that, owing to an admixture of some lighter molecules, the density would be somewhat reduced. But two other suppositions were entertained as possible. The oxygen might have been imperfectly removed from the nitrogen derived from the atmosphere; or, on the other hand, the nitrogen from ammonia might conceivably have retained traces of hydrogen. In the former case, the nitrogen would have an increased weight owing to admixture of some heavier oxygen; in the latter, a diminished weight, due to the presence of the lighter hydrogen. The first of these suppositions is out of the question, inasmuch as it would have required that the nitrogen should contain one-thirtieth of its volume of oxygen, or one-sixth of that present in air, in order that its density should be raised by one two-hundredth; for the densities of oxygen and nitrogen are not so very different. The second supposition was negatived by introducing hydrogen purposely, and removing it by passing the gas over red-hot copper oxide, which oxidises the hydrogen to water. This yielded nitrogen of the same density as that which had not undergone that treatment. One other possibility was considered: the atmospheric nitrogen might contain some molecules of greater complexity than two-atom molecules, say N_{3}-molecules. Now it is known that when oxygen is electrified by the passage of a rain of small sparks through it, it acquires new properties: it possesses an odour, and attacks metallic mercury and silver. And this product, ozone, has been shown to consist of three-atom molecules of oxygen, by various experiments of which an account cannot be given here.