180 Proceedings of Royal Society of Eclinhurgh. [sess. 
dry by sulphuric acid, until the surface was free from deposit of 
moisture and the tubes had taken the temperature of the room. 
The tubes were then reweighed. 
Weighings were made to milligramme ; but the impossibility 
of finding a perfectly satisfactory place for the balance made the 
conditions of weighing not quite ideal. 
The average change of weight in the standard tube, which had 
no air drawn through during these experiments, was from +0-53 to 
- 0*57 milligrammes, and that of the safety tube, omitting the cases 
where the first tube obviously did not dry the air, was from + 0*55 
to - 0*52 milligrammes. The latter figures represent an average 
gain and also an average loss of less than 0*04 milligrammes per 
litre of air aspirated. 
Results , — Nearly a hundred experiments were carried out at the 
High-Level Observatory, and of these at least half a dozen are 
valueless owing to breakages, and a score were made in a super- 
saturated atmosphere. The records of fully sixty experiments are 
available. From these a diagram has been drawn in which the 
abscissae represent temperatures of the dry-bulb thermometer, the 
ordinates the difference between the temperatures of dry and wet 
bulb thermometers, and the amount of water is noted for each 
experiment. On examining the figure it is found that three straight 
lines can be drawn through points representing 0’5, LO, and 2’0 
milligrammes of water vapour per litre of air. These are approxi- 
mately parallel and equidistant. Sufficient data have not yet been 
obtained to allow these lines to be drawn more precisely, but experi- 
ments carried out in summer should give the results needed to extend 
the present lines and to show their exact form, as well as to increase 
the number of lines. 
Assuming these lines to be parallel and equidistant, their equation 
is (when t is the temperature of the dry-bulb thermometer in centi- 
grade degrees, t-t' the difference between it and that of the wet- 
bulb one, and w is the weight of water vapour per litre of air 
•expressed in milligrammes) 
t-t' = ^ft -f 20) - 1 • . 
About twenty-five per cent, of the data used in preparing the dia- 
gram do not agree with the lines drawn from this equation. The most 
