PROFESSOR THOMAS GRAHAM's SCIENTIFIC WORK. 191 



But of all tbe results obtained, tlie most interestin.cf, from tlieir bear- 

 ing on various natural phenomena, were tbose on the partial separa- 

 tion of different compounds from one another, brought about by their 

 unequal diffusibility. Thus, with a solution of equal weights of com- 

 mon salt and gum-arabic placed in the diffusion-^ial, for every 100 milli- 

 grams of salt, not more than 22.5 milligrams of gum were found to 

 pass into the external water ; or a separation of the salt from the gum, 

 to this large extent, took i)lace spontaneously by the excess of its own 

 proper diffusive movement. Again, when a solution, containing 5 

 l)er cent, of common salt and 5 per cent, of Glauber's salt, was sub- 

 mitted for seven days to tlie process of jar-diffusion, the upper half, or 

 y^j, of superincumbent water was found to contain 3S0 milligrams of 

 common salt and only 53 milligrams of Glaul)er's salt ; or the ratio of 

 common salt to Glauber's salt in the upper half of the liquid was as 100 

 to 11, the ratio in the original stratum of solution being as 100 to 100. 

 And not onl^^ a partial separation of mixed salts, but even a partial 

 decomposition of chemical compounds, was found to result from the pro- 

 cess of liquid diffusion. Thus the double sulphate of potassium and 

 hydrogen, when submitted to diffusion, underwent partial decomposi- 

 tion into the more diffusible sulphate of hydrogen and the less ditfusible 

 sulphate of potassium; and, similarly, ordinary alum, a double sulphate 

 of aluminum and potassium, underwent partial decomposition into the 

 more diffusible sulphate of potassium, and the less diffusible sulphate of 

 aluminum. Strictly speaking, perhaps, the decomposition of the 

 original salts was not caused by, but only made evident by, the differ- 

 ence in diffusibility of the products. 



As a general result of his experiments, Mr. Graham inferred that 

 liquid diffusibility is not associated in any definite way with chemical 

 composition or molecular weight. Thus he found the complex organic 

 bodies picric acid and sugar to have much the same diffusive rates as 

 common salt and Epsom salt respectively. Isomorphous compounds, 

 however, proved for the most part to be equi-diffusive; although the 

 groups of equi-diffusive substances habitually comprehended other than 

 those which were isomorphous. 



Observing further that, in many cases, the diffusion-rates of different 

 equi-diffusive groups stood to one another in some simple numerical 

 relation, Mr. Graham remarked that, "In liquid diffusion we no longer 

 deal with chemical equivalents or the Daltonian atoms ; but with masses 

 even more simply related to each other by weight." We may suppose 

 that the chemical atoms "group together in such numbers as to form 

 new and larger molecules of equal weights for different substances, 

 Qj. * * * q£ weights which appear to have a simple relation to each 

 other;" and he inferred that the relative weights of these new molecules 

 would be inversely as the square roots of the observed diffusion rates of 

 the substances — that is inversely as the squares of their diffusion times. 

 Thus the squares of the times of equal diffusion of hydrate, nitrate, and 



