Graham divided substances into those which diffused easily and quickly into water, and those which diffused very slowly; he showed that the former were all crystallizable substances, while the latter were non-crystallizable jelly-like bodies. Graham called these jelly-like substances _colloids;_ the easily diffusible substances he called _crystalloids_. He proved that a colloidal substance acts towards a crystalloid much as water does; that the crystalloid rapidly diffuses through the colloid, but that colloids are not themselves capable of diffusing through other colloids. On this fact was founded Graham's process of _dialysis_. As colloid he employed a sheet of parchment paper, which he stretched on a ring of wood or caoutchouc, and floated the apparatus so constructed--_the dialyser_--on the surface of pure water in a glass dish; he then poured into the dialyser the mixture of substances which it was desired to separate. Let us suppose that this mixture contained sugar and gum; the crystalloidal sugar soon passed through the parchment paper, and was found in the water outside, but the colloidal gum remained in the dialyser. If the mixture in the dialyser contained two crystalloids, the greater part of the more diffusible of these passed through the parchment in a short time along with only a little of the less diffusible; a partial separation was thus effected. This method of dialysis was applied by Graham to separate and obtain in the pure state many colloidal modifications of chemical compounds, such as aluminium and tin hydrates, etc. By his study of these peculiar substances Graham introduced into chemistry a new class of bodies, and opened up great fields of research. Matter in the colloidal state appears to be endowed with properties which are quite absent, or are hidden, when it is in the ordinary crystalloidal condition. Colloids are readily affected by the smallest changes in external conditions; they are eminently unstable bodies; they are, Graham said, always on the verge of an impending change, and minute disturbances in the surrounding conditions may precipitate this change at any moment. Crystalloids, on the other hand, are stable; they have definite properties, which are not changed without simultaneous large changes in surrounding conditions. But although, to use Graham's words, these classes of bodies "appear like different worlds of matter," there is yet no marked separating line between them. Ice is a substance which under ordinary conditions exhibits all the properties of crystalloids, but ice formed in contact with water just at the freezing point is not unlike a mass of partly dried gum; it shows no crystalline structure, but it may be rent and split like a lump of glue, and, like glue, the broken pieces may be pressed together again and caused to adhere into one mass.