In order better to follow the train of reasoning based on experiment, it will be well to begin with an account of why the atomic weight of mercury is accepted as 200. The amount of mercury which combines with 16 parts by weight of oxygen is easily found by heating a weighed quantity of oxide of mercury, as Priestley and Scheele did, and weighing the residue of metal. The results of the most accurate experiments show that 200·36 grams of mercury combine with 16 grams of oxygen, and if the compound consists of one atom of each element, 200·36 must be the atomic weight of mercury. The first idea which naturally occurs is to find out the relative weight of mercury gas. This has been done, and it is found to have the ratio to that of oxygen of 100 to 16. Doubling these numbers will give the molecular weights, since a molecule of oxygen consists of two atoms, and must therefore possess twice the weight of one atom. We thus obtain the numbers 200 and 32 as the molecular weights of mercury and oxygen respectively. It might therefore be concluded that 200 is not the true atomic weight of mercury, but 100, and that the compound of mercury with oxygen contains not one but two atoms of mercury, and should therefore be represented by the formula Hg_{2}O, not HgO. But on surveying all known compounds of mercury, there is not one which contains less than 200 parts by weight of mercury in a molecule of the compound, or in which the mercury cannot be conceived to replace 2 parts by weight of hydrogen. And on weighing as gases the compounds of mercury with other elements, where such compounds do not decompose on heating like the oxide, the amount of mercury present must be always taken as 200, in order to add up to the molecular weight found. For example, a compound of mercury with carbon and hydrogen, named mercury methide, has a density of 120 compared with oxygen taken as 16, hence the comparative weight of its molecule must be 240. Now it is known to contain two atoms of carbon and six atoms of hydrogen, the atomic weights of which are 24 + 6 = 30. And deducting 30 from 240, 210 remains as an approximation to the atomic weight of mercury. It might, it is true, be the weight of two atoms of mercury, but if so it is singular that no compound contains a smaller proportion; and there is another reason, which follows immediately, that leads us to believe that 200 is correctly taken as the true weight of an atom.