But there still remained, in Cavendish’s opinion, one point unproved. It was still conceivable that the potash might contain some “inflammable matter” which would diminish the air on sparking, and therefore oxygen nearly pure was sparked in presence of potash; but only a very small diminution of volume occurred, owing probably to some nitrogen present as an impurity in the oxygen. Water was substituted for potash with the same result; but if litmus was added to the water the colour was discharged, and lime-water introduced into the tube gave a cloud, showing that “the litmus, if not burnt, was at least decompounded, so as to lose entirely its purple colour and to yield fixed air; so that, though soap-lees cannot be decompounded by the process, yet the solution of litmus can, and so very likely might the solutions of many other combustible substances.” Such are the chemical researches of Cavendish. Of all experimenters on the subject he was undoubtedly the greatest, though Mayow and Scheele were near rivals. But his researches were so complete that it is scarcely possible to criticise. He was not content with partial results: every point was proved and re-proved, and every possibility of erroneous conclusion was allowed for. It is curious that he did not employ the balance to check his results. Had he done so he could not have remained an adherent of the phlogistic theory. Although, as we have seen, he was perfectly acquainted with the method in which his results were interpreted by Lavoisier, he chose the old well-trodden path leading to the wilderness of distorted facts. Lavoisier tried to repeat Cavendish’s experiments, but without success; and in 1788 the last part of his _Experiments on Air_ was published, in which he recorded the successful repetition by a Committee of the Royal Society of the conversion of nitrogen into nitric acid by the electric spark in presence of oxygen and potash. His remaining papers deal with meteorological and astronomical subjects. One, published in 1790, refers to the height of a remarkable aurora seen in 1784; another to the civil year of the Hindoos; and another to a method for reducing lunar distances. And in 1798 his famous memoir on the density of the earth appeared. It would be quite beyond the province of this book to enter into any detail regarding it; but it may be remarked in passing that the method consisted in measuring, by means of a torsion balance, the attraction of one leaden ball for another, and that recent experiments, made with the utmost refinement, have barely altered the number which he obtained, 5·4, to 5·527. His last paper, on an improvement in a machine for dividing astronomical instruments, was published in 1809, the year before his death.