I must acknowledge myself not quite satisfied with Berthollet's reasoning on this subject. No doubt if two atoms of a body having a weaker affinity, and one atom of a body having a stronger affinity, were placed at equal distances from an atom of a third body, the force of the two atoms might overcome that of the one atom. And it is possible that such cases may occasionally occur: but such a balance of distances must be rare and accidental. I cannot but think that all the cases adduced by Berthollet are of a complicated nature, and admit of an explanation independent of the efficacy of mass. And at any rate, abundance of instances might be stated, in which mass appears to have no preponderating effect whatever. Chemical decomposition is a phenomenon of so complicated a nature, that it is more than doubtful whether we are yet in possession of data sufficient to enable us to analyze the process with accuracy. Another opinion brought forward by Berthollet in his work was of a startling nature, and occasioned a controversy between him and Proust which was carried on for some years with great spirit, but with perfect decorum and good manners on both sides. Berthollet affirmed that bodies were capable of uniting with each other in all possible proportions, and that there is no such thing as a definite compound, unless it has been produced by some accidental circumstances, as insolubility, volatility, &c. Thus every metal is capable of uniting with all possible doses of oxygen. So that instead of one or two oxides of every metal, an infinite number of oxides of each metal exist. Proust affirmed that all compounds are definite. Iron, says he, unites with oxygen only in two proportions; we have either a compound of 3·5 iron and 1 oxygen, or of 3·5 iron and 1·5 oxygen. The first constitutes the _black_, and the second the _red_ oxide of iron; and beside these there is no other. Every one is now satisfied that Proust's view of the subject was correct, and Berthollet's erroneous. But a better opportunity will occur hereafter to explain this subject, or at least to give the information respecting it which we at present possess.