824 
Proceedings of the Royal Society 
solutions which contain the same percentage of different salts are 
not comparable. Before we can justly expect to he able to throw 
light on the subject of solution, we must he certain that the 
solutions which we compare together are really comparable. 
Kremers, * in the course of his numerous experiments on solution, 
and his deductions from them, compared solutions which contained, 
as he expressed it, 10, 20, 30 molecules of the salt ( Salzatome ) to 
the 100 unit weights or unit volumes of water ( Gemchtseinheiten 
oder Volumeinheiten Wasser ): these numbers, as a reference to his 
papers will show, are too great, hut for all that his solutions were to 
a certain extent comparable. The source of error which Kremers 
did not perceive, which indeed so far as I know has not hitherto 
been recognised, is the mutual attraction of the salt molecules for one 
another. If we wish to observe the results produced solely by the 
mutual action of the salt and the water, we must take care that our 
solutions are sufficiently dilute. As the solution becomes more 
concentrated, the observed specific gravity depends not only on the 
attraction of the salt for the water, hut also on that of the salt 
molecules for one another. 
With a view to testing the correctness of the above, I plotted the 
results given in Table II. cols. C. and D. The ordinates representing 
the specific gravity and the abscissae the percentages ; I thus obtained 
for each salt two curves approaching straight lines. Both turned 
upwards, but that expressing the observed specific gravity turned 
less than the other. The curvature in either case was so slight that 
it was impossible to draw with any approach to accuracy curves 
joining the various points; these were therefore joined by straight 
lines ; from these lines the specific gravity for any percentage can 
be found with moderate exactness ; but the difference between the 
observed and calculated specific gravities can be found with 
much greater accuracy, as the error affects both lines to very 
nearly the same extent. By the help of these lines, the actual 
and theoretical specific gravities were found for solutions con- 
taining 1, 2, 4, 6, and even 8 molecules of salt to 100 mole- 
cules of water. The differences between these are given in Table 
III. 
* Pogg., Annalen, vols. xcii., xcv., cxvi., xcviii.,&c. 
