282 



Prof. W. J. Sollas. 



As close a correspondence appears as could be expected from the 

 conditions of the case, and it would seem that the crystalline struc- 

 ture attributed to the haloid salts, while inconsistent with no known 

 group of facts, is in quantitative accordance with all that we have 

 investigated, and throws unexpected light on hitherto recondite 

 phenomena. The subject of solution must not, however, be left 

 without giving the densities found for more dilute solutions : they 

 are shown in the following table : 



Some of these numbers are less concordant with the density 

 deduced from the crystalline structure than those obtained with 

 stronger solutions, as presented in the preceding table, but it is to be 

 remarked that whatever errors exist in the observations are all 

 thrown on to the values we have obtained ; the numbers given in the 

 third column bear all the burden of error arising when dealing wifch 

 very minute quantities. To show how greatly our results are influ- 

 enced by slight variations in specific gravity, as given in the second 

 column of numbers above, I have calculated backwards from the 

 density of the atoms to find what specific gravity should give us 

 identical values for the density, as calculated from the crystalline 

 structure and from solution. It is not necessary to give the results 

 for more than two salts ; in the case of potassium chloride the 

 specific gravity of the solution, to accord with theory, should be 

 1-0038 instead of 1*0046 observed, to reduce the density 3 -9989 to 

 2*7668, its value on the assumptions we have made; in the case of 

 sodium chloride the specific gravity should be 1*0019 to give a 

 density of 3*0467, instead of the 1*0038 found. Although different 

 observers do not always give the same specific gravity for the same 

 solutions of salts, it is certain that the discrepancies between theory 

 and observation before us are not to be explained away by blaming 

 the observations ; we have before us an interesting residual pheno- 

 menon, susceptible of more explanations than one. 



It will be observed that those salts the density of which is in 

 excess of that predicted, are those whose volumes are relatively 



