In 1836 Graham's paper on "Water as a Constituent of Salts" was published in the "Transactions of the Royal Society of Edinburgh." In this paper he inquires whether the water in crystalline salts can or cannot be removed without destroying the chemical individuality of the salts. He finds that in some crystalline salts part of the water can be easily removed by the application of heat, but the remainder only at very high temperatures. He distinguishes between those atoms of water which essentially belong to the compound atom of the salt, and those atoms which can be readily removed therefrom, which are as it were added on to, or built up around the exterior of the atom of salt. In this paper Graham began to distinguish what is now called _water of crystallization_ from _water of constitution_, a distinction pointed to by some of Davy's researches, but a distinction which has remained too much a mere matter of nomenclature since the days of Graham. In these researches Graham emphasized the necessity of the presence of hydrogen in all true acids; as he had drawn an analogy between water and bases, so now he saw in the hydrogen of acids the analogue of the metal of salts. He regarded the structure of the compound atom of an acid as similar to that of the compound atom of a salt; the hydrogen atom, or atoms, in the acid was replaced by a metallic atom, or atoms, and so a compound atom of the salt was produced. Davy and Berzelius had proved that hydrogen is markedly electro-positive; hydrogen appeared to Graham to belong to the class of metals. In making this bold hypothesis Graham necessarily paid little heed to those properties of metals which appeal to the senses of the observer. Metals, as a class, are lustrous, heavy, malleable substances; hydrogen is a colourless, inodourless, invisible, very light gas: how then can hydrogen be said to be metallic? I have again and again insisted on the need of imagination for the successful study of natural science. Although in science we deal with phenomena which we wish to measure and weigh and record in definite and precise language, yet he only is the successful student of science who can penetrate beneath the surface of things, who can form mental pictures different from those which appear before his bodily eye, and so can discern the intricate and apparently irregular analogies which explain the phenomena he is set to study.