Dumas's experiments were summed up in the statement that the atom of carbon is twelve times heavier than the atom of hydrogen. The experimental methods and the calculations used in this determination involved a knowledge of the atomic weight of oxygen; in order accurately to determine the value to be assigned to this constant, Dumas, in conjunction with Boussingault, undertook a series of experiments on the synthesis of water, which forms one of the classical researches of chemistry, and wherein the number 16 was established as representing the atomic weight of oxygen. Stas, from experiments conducted at a later time with the utmost care and under conditions eminently fitted to gain accurate results, obtained the number 15.96, in place of 16, for the atomic weight of oxygen; but in a paper recently published by the veteran Dumas, a source of error is pointed out which Stas had overlooked in his experiments, and it is shown that this error would tend slightly to increase the number obtained by Stas. As the values assigned to the atomic weights of the elements are the very fundamental data of chemistry, and as we are every day more clearly perceiving that the mutual relations between the properties of elements and compounds are closely connected with the relative weights of the elementary atoms, we can scarcely lay too much stress on such work as this done by Dumas and Stas. Not many years after the publication of Dalton's "New System," the hypothesis was suggested by Prout that the atomic weights of all the elements are represented by whole numbers--that of hydrogen being taken as unity--that the atom of each element is probably formed by the putting together of two, three, four, or more atoms of hydrogen, and that consequently there exists but a single elementary form of matter. Among the upholders of this hypothesis Dumas has held an important place. He modified the original statement of Prout, and suggested that all atomic weights are whole multiples of half of that of hydrogen (that is, are whole multiples of 1/2). The experiments of Stas seemed to negative this view, but later work--more especially the important critical revision of the results obtained by all the most trustworthy workers, conducted by Professor Clarke of Cincinnati, and published by the Smithsonian Institution as part of their series of "Constants of Nature"--has shown that we are in no wise warranted by facts in rejecting Prout's hypothesis as modified by Dumas, but that the balance of evidence is at present rather in its favour.