360 Mr. S. U. Pickering on the Densities 



solutions are very numerous, and if it is established that the 

 properties of the solutions show changes of curvature at points 

 corresponding to these isolated hydrates, we shall have further 

 evidence of a very strong character in favour of the reality of 

 these changes, and of the view that they indicate the existence 

 of hydrates in solution. 



The Determinations. 



The soda used was made from sodium, and the potash was 

 purified by crystallization from alcohol. The lumps were first 

 washed in well-boiled water to remove any adhering carbonate, 

 and were then dissolved to form solutions nearly saturated 

 at ordinary temperatures, and filtered through glass-wool. 

 These strong solutions were subsequently diluted to the required 

 strengths by the addition of accurately weighed quantities 

 of water. The solutions were made up twelve to twenty-four 

 hours before their densities were determined, specially stopped 

 bottles of common white glass being used for holding them 

 during this time. The same solution was never used twice. 

 In spite of every precaution the solutions could not be obtained 

 entirely free from carbonate : the presence of a little car- 

 bonate, however, would have no appreciable effect on the 

 relative accuracy of consecutive observations, which is the 

 chief consideration in the present work. 



The strength of the stock solutions was determined by 

 titration, and was probably correct to 0'05 per cent, of the total 

 alkali present. The relative strength of the consecutive 

 solutions is dependent only on the balance error in diluting 

 them, and is therefore represented by a much smaller 

 quantity. 



The density-determinations were made in a precisely 

 similar manner as in the case of sulphuric acid, the tempera- 

 ture, however, being 15° + *01. The various determinations 

 of consecutive strengths were not always made consecutively. 

 The results with soda, for instance, were made in five or six 

 different series, each extending throughout the whole range of 

 strengths investigated, and any changes which are noticed 

 cannot, therefore, be explained aw 7 ay by the suggestion of 

 there being a certain error affecting a certain day's work, and 

 not affecting that of another day. 



