544 MR J. V. BUCHANAN ON 



KC1 present in greater quantity than could in any length of time be dissolved in the 

 amount of water present. Again, if we look at experiment No. 4, we find the observed 

 value of Wj, 27'25, almost identical with the computed value of W , 27 '"21. Here the 

 difference, t — t 1 , is - 2° C. So that, working in the way described, with chloride of 

 potassium, when the temperature of condensation of the steam has fallen by 0*2° C, we 

 should expect to have the amounts of free water and free salt present in compensating 

 amounts. No. 6 represents a case in which the steam was passed until all the salt was 

 dissolved, and until the temperature of condensation had fallen by a whole degree. 

 Here we use W 1 (t-T) 1 and W 2 (*-T) 2 for finding W (*-T) and W . If we con- 

 sider experiments Nos. 1 to 9, it will be seen that the values of W in experiments 1 to 

 3, which were made in Edinburgh, are lower than those found in Nos. 4 to 9, which 

 were made at high levels, and therefore at lower temperatures, in Switzerland. 



It must also be noted that the weight of the salt taken was less exactly ascertained 

 in Switzerland than in my laboratory in Edinburgh. The Swiss weighings were made 

 with a pair of hand scales, and were exact to the nearest 0'05 grm., that is to say, 

 generally to ±0"025 grm. In the Edinburgh experiments quoted in Table IV. the 

 weights are exact to the nearest 0*01 grm., or to ±0*005 grm. The quantity of salt 

 usually taken was one-fifth of a molecule in grammes. In the case of KC1, which has a 

 medium molecular weight, this represents 14 - 92 grms., and we see that even the roughest 

 of the weighings would be exact to within less than one-half per cent. 



The earlier experiments in Switzerland were not always made with equivalent 

 weights of the salts ; all such cases have been recalculated for this table. While the 

 usual quantity taken is one-fifth of a gramme-molecule, on some occasions two-fifths 

 have been taken ; and in the case of sparingly soluble salts as little as one-tenth or one- 

 twentieth has been taken. The values of W are given for the quantity of salt quoted 

 in the headline (M). The products have been all reduced to their value for one-fifth 

 of a molecule salt. 



The physical meaning of the expression W(t — T) is important. W is a weight of water 

 expressed in grammes, and (t — T) is the excess of the boiling temperature in degrees Cel- 

 sius of that water, when it holds in solution a certain amount of a given salt, above its boil- 

 ing temperature when in a state of purity ; therefore W(t — T) expresses, in gramme- 

 degrees (g° C), the quantity of heat required to be in the water when it is boiling with salt 

 dissolved in it above what is required when it is pure. If the values of W(£ — T) were 

 eonstant for each salt at all dilutions, then the law connecting the dilution of a saline 

 solution and the elevation of its boiling point would be graphically expressed by a hyper- 

 bola, like the law connecting the volume and pressure of a gas at constant temperature. 

 If we look over Table IV. we sec that for some salts, and mixtures of salts, the values 

 of W(j5 — T) are very nearly quite constant, while for the others, some deviate from con- 

 stancy in the one sense, and some in the other. In the case of the chlorides of potassium 

 and of sodium, the values of W(t — T) diminish very considerably as the value of W 

 increases. The case of ammonium chloride is peculiar, because at the sea level W(^-T) 



