I04 



NA TURE 



[December i, 1892 



posed. For suppose we have a quantity of solution 

 enclosed in a tube, one end of the tube being closed by a 

 membrane A, the other by a membrane B, and suppose 

 it possible that a pressure P can be developed on the 

 membrane A when it separates the solution frpm pure 

 water, which is higher than the pressure^ similarly de- 

 veloped when B separates the solution from pure water. 

 On immersing the tube in water, the latter will begin to 

 pass through both membranes into the solution. When 

 the pressure p is attained passage through B will stop, 

 but that through A will continue ; but as soon as the 

 pressure on the solution rises above ^, water will be forced 

 out through B. The pressure P will thus never be 

 attained, water will continuously enter through A, and 

 pass out at B. We will thus have a machine capable 

 of doing an infinite amount of work, which is impossible. 

 Similar reasoning shows that/ cannot be greater than P ; 

 it follows therefore that the pressure developed on each 

 membrane is the same, that the osmotic pressure must be 

 independent of the nature of a truly semi-permeable 

 membrane. 



Actual observations are on record in which the 

 osmotic pressure did appear to vary with the membrane 

 employed. A sugar solution, for example, exhibited a 

 much lower osmotic pressure with a membrane of Prussian 

 blue or calcium phosphate than with copper ferrocyanide. 

 From the preceding argument it is concluded, however, 

 that these membranes giving the lower values were not 

 quite firm or not quite impermeable to the dissolved 

 substance ; the highest value is thus taken as the measure 

 of the osmotic pressure which is nearest the truth. 



On glancing at the results which have been obtained, the 

 first point which strikes one is the extraordinary magni- 

 tude of the pressures thus set up. In the case of a i per 

 cent, aqueous solution of nitre the pressure attains the 

 value of 2i atmospheres. This value increases with the 

 strength of the solution till at 3'3 per cent, it is no less 

 than 6 atmospheres, this pressure being the highest which 

 any membrane yet prepared has been able to withstand. 

 With substances like sugar, other things being the same, 

 the pressure is not so great, but in all cases, in order to 

 keep it within workable limits, the solutions employed 

 have to be dilute. 



Striking as the results are themselves, their explanation 

 is not less remarkable. The original measurements of 

 osmotic pressure were made with the purpose of eluci- 

 dating the movement of liquids in plant cells, and 

 naturally the substances examined were such as occur 

 in the vegetable organism — aqueous solutions of sugar, 

 gum, dextrin, and the nitrate, sulphate, and tartrate 

 of potassium. For some years after these observations 

 were made, they lay comparatively unnoticed, until Prof, 

 van't Hoff, of Amsterdam, turned them to a use undreamt 

 of by their discoverer. From a study of the properties of 

 dilute solutions van't Hoff came to the conclusion that 

 the osmotic pressure was due to the bombardment of the 

 molecules of the dissolved substance on the semi-perme- 

 able membrane. For when the osmotic pressure is 

 established and equilibrium exists between solvent and 

 solution, in the same time, equal amounts of solvent, 

 must pass in either direction through the membrane 

 and the impacts of the solvent molecules on the mem- 

 brane will then be equal and opposed on either side, and 

 therefore negligible. On this reasoning the pressure 

 recorded on the manometer is taken to be that exerted 

 by the substance in solution. 



On examining the magnitude of the pressure thus at- 

 tributed to the dissolved substance, in the case of a solu- 

 tion of sugar van't Hoff next showed that it bore the 

 closest resemblance to the pressure of a gas. Indeed, if 

 we calculate the pressure of a gas which at the same 

 temperature contains as many molecules per unit volume 

 as there are molecules of sugar per unit volume of solu- 

 tion, then the pressure of the gas and the osmotic pres- 



NO. I205,|:VOL. 47] 



sure are the same. Moreover, on thermodynamical 

 grounds it was established that on the above hypothesis 

 as to the nature of osmotic pressure its magnitude should 

 be quantitatively connected with measurements of other 

 physical properties of solutions, more especially those on 

 the lowering of the vapour-pressure, and of the freezing 

 point of a solvent produced by the presence of dissolved 

 material. In this way a mass of evidence was collected, 

 a general survey of which led to the foundation of the 

 new theory of solutions. On this theory the dissolved 

 substance, if the solution be dilute, is supposed to behave 

 as if it were gaseous, the pressure it exerts — the osmotic 

 pressure — being equal to the pressure which it would exert 

 if it were gasified, and occupying, at the same tempera- 

 ture, a volume equal to the volume of the solution. 



Unfortunately measurements of osmotic pressure have 

 only been made on few substances, and only for solutions 

 in water, but on turning to all the available observations 

 to see how they support this novel conclusion, the most 

 superficial examination serves to show that an agreement 

 does not exist. Unless in the case of sugar, for no sub- 

 stance of known formula which has yet been investigated 

 does the osmotic pressure agree with the corresponding 

 gaseous pressure. These substances consist of salt solu- 

 tions, and they invariably give higher osmotic pressures 

 than theory demands. Similar disturbing influences 

 have been observed when other physical properties of 

 these solutions were measured, and to account for the 

 facts an additional hypothesis has been put forward by 

 Dr. Svante Arrhenius. 



Salt solutions are electrolytes, they conduct the electric 

 current, and undergo simultaneous chemical decomposi- 

 tion into their constituent ions. Experiment shows that 

 such electrolytic solutions give high osmotic pressures, 

 more particles appear to bombard the semi-permeable 

 membrane than if the dissolved substance behaved as a 

 gas. The new hypothesis states that this is really the 

 case, the additional number of particles being produced 

 from the breaking up of the dissolved substance. It 

 states that in a solution which can be electrolyzed a por- 

 tion at least of the dissolved substance exists already 

 decomposed or dissociated into its ions, and that although 

 these ions cannot be separated by diffusion they are sO' 

 far independent that each can exercise an effect on the 

 i semi-permeable membrane. 



The extent of this electrolytic dissociation is sup- 

 posed to vary with the chemical nature of the dissolved 

 substance, and to increase with the dilution. In very 

 dilute solutions it may be complete, the whole of the dis- 

 solved substance being supposed to exist in the state of 

 ions. 



The second hypothesis gives, therefore, some expla- 

 nation why the osmotic pressure of a salt solution is 

 greater than that of a non-electrolytic solution of sugar ; 

 it further fixes the limits between which the osmotic pres- 

 sure ought to vary in the case of an electrolyte, for the 

 lower limit should be that of undissociatedgas, the higher 

 should be that of completely dissociated gas, each 

 original molecule having decomposed into as many sub- 

 molecules as there are ions in each molecule of salt. 



So far as these limiting conditions go, the facts sup 

 port the hypothesis. In all cases the observed osmotic 

 pressure is either equal to one or other of the limits, or 

 lies between them. A closer scrutiny leads, nevertheless,, 

 to apparent discrepancy. It is evident that a measure of 

 the amount of dissociation can be obtained from osmotic 

 pressure observations. For if we divide the observed 

 osmotic pressure by the corresponding pressure of un- 

 dissociated gas we have obviously, if the preceding 

 hypotheses are valid, the ratio of the actual number of 

 bombarding molecules to the theoretical number had no- 

 dissociation occurred. The ratio of these two numbers 

 is denoted by the letter " i" a factor first used by van't 

 Hoff. Now, on the new theory, the value of " z" can be 



