24 
PROFESSOR W. A. TILDEN AND MR. W. A. SHENSTONE ON THE 
thermometer at each end, and a thermostat. The tube and its contents were then 
exposed to the desired temperature for a considerable time, usually about four-and-a- 
half to five hours, so as to ensure complete saturation. By then gradually tilting 
the air bath, which was slung upon suitable supports, the solution was caused to drain 
away from the undissolved residue of the salt, and the whole was then allowed to cool. 
Finally the tube was cut open, the end containing the solution was weighed, and. 
after removing the solution to a tared dish, was reweighed. The solution was then 
evaporated to dryness with due precautions, and the weight of the dry residue 
determined. From the data so obtained the proportion of anhydrous salt to water 
in the solution was calculated. 
Throughout these operations two tubes were invariably employed in each 
experiment, and two determinations thus made simultaneously. The mean of the 
results was taken. Preliminary experiments were also made which indicated that 
by a proper disposition of the four Bunsen burners used as the source of heat, and 
the employment of the mercurial gas regulator, the temperature of the air bath could 
be maintained within a range of fi; 2° for many hours. We also satisfied ourselves of 
the efficiency of the platinum gauze strainers, by frequently examining the decanted 
solution whilst still warm and liquid. It may also be added that the salts operated 
upon were pure and were almost always specially prepared in the form of agglom¬ 
erated masses of small crystals, with the object alike of exposing a greater surface to 
the action of the solvent, and of allowing the solution to flow away freely when the 
tube was reversed. 
The employment of bent glass tubes was soon abandoned, owing to the difficulty 
of keeping both ends of the tube at precisely the same temperature, also in considera¬ 
tion of the greater risk of bursting to which they were liable. The employment of 
tubes of glass under any form is, in the case of sulphate of sodium, undesirable, as we 
find that this salt has the power of attacking glass at high temperatures in an 
unexpected degree. Consequently, although we succeeded in satisfying ourselves 
that the solubility of sodium sulphate in water at temperatures above the normal 
boiling point of the solution does increase in the manner w r e anticipated, the 
numerical results of the successive experiments were not sufficiently concordant to 
show clearly the form of the continuation of Gay Lussac’s curve. It therefore 
became necessary to employ tubes of metal, and of a somewhat different construction. 
In arranging the experiments, which were afterwards extended to many other 
metallic salts, the following considerations had to be taken into account :— 
1. It is necessary to allow the solution to drain away completely from the undis¬ 
solved residue of salt before allowing the tube to cool; otherwise a part of the 
dissolved salt may be deposited, and the solution collected for analysis would be 
weaker than it should be. Provision is made against this source of error in the metal 
tube we have used in the later experiments. 
2. If the two extremities of the tube are not kept at absolutely equal temperatures, 
