+--------------+-------------+---------------+ | Decrease of | Cube root of| Ratio between | | solubility | molecular | decrease and | | per 20° in | weight. | cube root of | | per cent. | | mol. wt. | +--------------+-------------+---------------+ | H_{2} 15·32 | 1·259 | 12·17 | | N_{2} 34·33 | 3·037 | 11·30 | | CO 34·44 | 3·037 | 11·34 | | NO 36·24 | 3·107 | 11·66 | | O_{2} 36·55 | 3·175 | 11·51 | +--------------+-------------+---------------+ The decrease in the coefficient of absorption with the temperature must be connected with a change in the physical properties of the water. Winkler (1891) remarked a certain relation between the internal friction and the coefficient of absorption at various temperatures. [31] These figures show that the co-efficient of solubility decreases with an increase of pressure, notwithstanding that the carbonic anhydride approaches a liquid state. As a matter of fact, liquefied carbonic anhydride does not intermix with water, and does not exhibit a rapid increase in solubility at its temperature of liquefaction. This indicates, in the first place, that solution does not consist in liquefaction, and in the second place that the solubility of a substance is determined by a peculiar attraction of water for the substance dissolving. Wroblewski even considered it possible to admit that a dissolved gas retains its properties as a gas. This he deduced from experiments, which showed that the rate of diffusion of gases in a solvent is, for gases of different densities, inversely proportional to the square roots of their densities, just as the velocities of gaseous molecules (see Note 34). Wroblewski showed the affinity of water, H_{2}O, for carbonic anhydride, CO_{2}, from the fact that on expanding moist compressed carbonic anhydride (compressed at 0° under a pressure of 10 atmospheres) he obtained (a fall in temperature takes place from the expansion) a very unstable definite crystalline compound, CO_{2} + 8H_{2}O. [32] As, according to the researches of Roscoe and his collaborators, ammonia exhibits a considerable deviation at low temperatures from the law of Henry and Dalton, whilst at 100° the deviation is small, it would appear that the dissociating influence of temperature affects all gaseous solutions; that is, at high temperatures, the solutions of all gases will follow the law, and at lower temperatures there will in all cases be a deviation from it.