He next directed his attention to the discovery of some method by which the _gallic acid_ might be separated from _extractive matter_, in cases where they exist in combination, but the enquiry was not successful; for, as he observes, it is difficult to render the _extractive_ insoluble, so as to separate it, without at the same time decomposing the gallic acid. It is true that æther will dissolve the latter, without exerting much action upon the former; but then, he adds, whenever the gallic acid is in large quantities, this method will fail, "in consequence of that _affinity_ which is connected with mass." Here then he adopts that celebrated theory of Berthollet,[42] which he afterwards so vigorously and successfully attacked.[43] [42] Récherches sur les Lois de l'Affinité.--Mém. de l'Institut National, Tome III. p. 5. [43] The masterly manner in which he combated the successive arguments of Berthollet upon this question is admirable. In the first place, he attacked the theory upon general principles, and then exposed the fallacy of the several experiments adduced in its support. "Were the proposition correct, that _in all cases of decomposition in which two bodies act upon a third, that third is divided between them in proportion to their relative affinities, and their quantities of matter_, it is quite evident that there could be scarcely any definite proportions: a salt crystallizing in a strong alkaline solution would be strongly alkaline; in a weak one, less alkaline; and in an acid solution it would be acid." With regard to glasses and alloys, adduced by M. Berthollet as compounds of indefinite proportions, Davy answers--"It is not easy to prove, in such cases, that the elements are chemically combined, for the points of fusion of alkali, glass, and certain metallic oxides, are so near to each other, that transparent mixtures of them may be formed." The experiment upon which M. Berthollet laid great stress, viz. that a large quantity of potash will separate a small quantity of sulphuric acid from sulphate of baryta, Davy invalidates in a most complete manner. He says--"This experiment was made in contact with the atmosphere, in which carbonic acid is always present; and carbonate of potash and sulphate of baryta mutually decompose each other." As general tests of the respective quantities of these two principles (gallic acid and extractive matter), he employed the solutions of the salts of alumina and those of the peroxidated salts of iron. The former of these precipitates _extractive_, without materially acting upon _gallic acid_, which is thrown down by the latter: the greatest care, however, must be taken not to add the iron in excess, as in that case the black precipitate formed will be redissolved, and an olive-coloured and clear fluid be only obtained.