4. This step accordingly was made, when the hypothesis of the four elements was given up, and the doctrine of the _three Principles_, Salt, Sulphur, and Mercury, was substituted in its place. For in making this change, as I have remarked in the History[5\6], the real advance was the acknowledgment of the changes, produced by the chemist's operations, as results to be accounted for by the union and separation of substantial elements, however great the changes, and however unlike the product might be to the materials. And this step once made, chemists went on constantly {7} advancing towards a truer view of the nature of an element, and consequently, towards a more satisfactory theory of chemical operations. [Note 5\6: _Hist. Ind. Sc._ b. iv. c. 1.] 5. Yet we may, I think, note one instance, even in the works of eminent modern chemists, in which this maxim, that we have no right to expect any resemblance between the elements and the compound, is lost sight of. I speak of certain classifications of mineral substances. Berzelius, in his System of Mineral Arrangement, places _sulphur_ next to the _sulphurets_. But surely this is an errour, involving the ancient assumption of the resemblance of elements and compounds; as if we were to expect the sulphurets to bear a resemblance to sulphur. All classifications are intended to bring together things resembling each other: the sulphurets of metals have certain general resemblances to each other which make them a tolerably distinct, well determined, class of bodies. But sulphur has no resemblances with these, and no analogies with them, either in physical or even in chemical properties. It is a simple body; and both its resemblances and its analogies direct us to place it along with other simple bodies, (selenium, and phosphorus,) which, united with metals, produce compounds not very different from the sulphurets. Sulphur cannot be, nor approach to being, a sulphuret; we must not confound what it _is_ with what it _makes_. Sulphur has its proper influence in determining the properties of the compound into which it enters; but it does not do this according to resemblance of qualities, or according to any principle which properly leads to propinquity in classification.