An indispensable condition of the circulation of matter (metabolism) which we call life is the physical process of osmosis, which is connected with the variations in the quantity of water in the living substance and its power of diffusion. The plasm, which is of a spongy or viscous consistency, can take in dissolved matter from without (endosmosis) and eject matter from within (exosmosis). This absorptive property (or "imbibition-energy") of the plasm is connected with the colloidal character of the albuminoids. As Graham has shown, we may divide all soluble substances into two groups in respect of their diosmosis--crystalloids and colloids. Crystalloids (such as soluble salt and sugar) pass more easily into water through a porous wall than colloids (such as albumen, glue, gum, caramel). Hence we can easily separate by dialysis two bodies of different groups which are mixed in a solution. For this we need a flat bottle with side walls of india-rubber and bottom of parchment. If we let this vessel float in a large one containing plenty of water, and pour a mixture of dissolved gum and sugar into the inner vessel, after a time nearly all the sugar passes through the parchment into the water, and an almost pure solution of gum remains in the bottle. This process of diffusion, or osmosis, plays a most important part in the life of all organisms; but it is by no means peculiar to the living substance, any more than the absorptive or viscous condition is. We may even have one and the same substance--either organic or inorganic--in both conditions, as crystal or as colloid. Albumen, which usually seems to be colloidal, forms hexagonal crystals in many plant-cells (for instance, in the aleuron-granules of the endosperm), and tetrahedric hœmoglobin-crystals in many animal-cells (as in the blood corpuscles of mammals). These albuminoid crystals are distinguished by their capacity for absorbing a considerable quantity of water without losing their shape. On the other hand, mineral silicon, which appears as quartz in an immense variety (more than one hundred and sixty) of crystalline forms, is capable in certain circumstances (as metasilicon) of becoming colloidal and forming jelly-like masses of glue. This fact is the more interesting because silicium behaves in other ways very like carbon, is quadrivalent like it, and forms very similar combinations. Amorphous (or non-crystalline) silicium (a brown powder) stands in relation to the black metallic silicon-crystals just as amorphous carbon does to graphite-crystals. There are other substances that may be either crystalloid or colloid in different circumstances. Hence, however important colloidal structure may be for the plasm and its metabolism, it can by no means be advanced as a distinctive feature of living matter.