ox WATKK AND ITS COMPOUNDS 91 



to the amounts of salt in solution (1, 5, and 10 per 100 of water). 

 Furthermore, it has been shown by experiment that the ratio of the 

 diminution of vapour tension to the vapour tension of water at different 

 temperatures in a given solution is an almost constant quantity/" and 



temperature of the formation of ice, of osmotic pressure (Van't Hoff, note 19), and of the 

 electrical conductivity <>f solutions, and we will therefore supplement what we have 

 ul ready said on the subject by some short remarks on the method of investigating the 

 phenomenon, and on its theretical results. 



In order to determine the temperature of the formation of ice (or of crystallisation 

 of other solvents), a known solution is prepared and poured into a cylindrical vessel 

 surrounded by a second similar vessel, leaving a layer of air between the two, which, 

 being a bad conductor, prevents any rapid change of temperature. The bulb of a sensi- 

 tive and corrected thermometer is immersed in the solution, and also a bent platinum 

 wire for stirring the solution ; the whole is then cooled (by immersing the apparatus in a 

 freezing mixture), and the temperature at which ice begins to separate observed. If the 

 temperature at first falls slightly lower, nevertheless, it becomes constant when ice 

 begins to form. By then allowing the liquid to get just warm, and then again observing 

 the temperature of the formation of ice, an exact determination may be arrived at. If 

 there be a large mass of solution, the formation of the first crystals may be accelerated 

 by dropping a small lump of ice into the solution already partially over-cooled. This 

 only imperceptibly changes the composition of the solution. The observation should be 

 made at the point of formation of only a very small amount of crystals, as otherwise the 

 composition of the solution will become altered from their separation. Every precaution 

 must be taken to prevent the access of moisture to the interior of the apparatus, which 

 might also alter the composition of the solution or properties of the solvent (for instance, 

 when using acetic acid). 



The very great theoretical interest of these observations on the fall of the tempera- 

 ture of the formation of ice, which are essentially very simple, dates from the time when 

 Van't Hoff (note 19) showed that their consequences are in complete accord with those 

 derived from observations on osmotic pressure. These latter showed that a molecular 

 (expressed by formulae) quantity of a substance evinces an osmotic pressure in a solu- 

 tion, which is equal to the atmospheric pressure (when i = 1), or which is greater than it 

 by i times. The magnitude i, determined from osmotic observations on aqueous solutions, 

 is also obtained from observations on the fall of the temperature of the formation of ice, 

 if the fall corresponding with a solution containing 1 gram of a substance per 100 parts 

 water be multiplied by the molecular weight (according to the formula of the substance, 

 and expressing the weight of a molecule) of the substance dissolved, and divided by 

 18'5. Thus from the above data for acetone, it is seen that with a solution containing 

 1 gram, the fall of temperature of the formation of ice equals 0'818, and after multiply- 

 ing by the molecular weight (58), and dividing by 18'5, we have i=l. With sugar and 

 many other substances (among salts, magnesium sulphate, for instance), with carbonic 

 anhydride, ttc., both methods give a figure which is nearly unity. For potassium and 

 sodium chlorides, potassium iodide, nitre, and others, i is greater than 1 but less than 

 2 ; for sulphuric and hydrochoric acids, sodium and calcium nitrates, and others, i is 

 nearly 2 ; for solutions of barium and magnesium chlorides, potassium carbonate and 

 dicliromate, i, according to both methods, is greater than 2 but less than 3. The further 

 investigation of this subject should show whether these conclusions are entirely general, 

 and would probably explain better than they do now those remarkable correlations 

 which are arrived at with the present data. 



M This fact, which was established by Gay-Lussac, Prinsep, and v. Babo, is confirmed 

 by the latest observations, and enables us to express not only the fall of tension (p p') 



its.!!, but its ratio to the tension of water (2.\. It is to be remarked that in the 



V p I 



absence of any chemical action, the fall of tension is either very small, or does not 



