6 
Proceedings of the Boyal Society of Edinburgh. [Sess. 
A slight increase in pressure can be produced in a by closing the top 
of the tube e with the finger and allowing e to communicate with a 
through the tap c. A slight decrease in pressure can be produced in a by 
closing the tap / and allowing the evacuated space between c and / to 
communicate with a. 
In this way any pressure up to one atmosphere can be obtained. By 
blowing air through e into the apparatus, readings may be made up to 
nearly 2 atmospheres. 
To test the efficiency of the method, the vapour density of mercury 
was determined. The accuracy of the determination depends necessarily 
on the accuracy with which the temperature of the vapour in the bulb A 
(fig. 3) can be measured. 
The bulb A was fitted with about 1 mm. play into the heating-tube of 
a Herseus electric furnace 60 cm. in length. A Le Chatelier thermo- 
couple tt was made to pass through the quartz tube L (fig. 3). L was 
sealed into the bulb A and passed through the interior of A, so that it 
was completely surrounded by mercury vapour. In this way the thermo- 
couple was prevented from coming into direct contact with the walls 
of the furnace, whose temperature often varies from point to point 
in an irregular manner. Those parts of the furnace wall which are 
covered with the heating spiral of platinum foil are generally at a con- 
siderably higher temperature than the immediately adjacent parts which 
are not covered by the heating foil. 
The thermocouple could be moved freely in L so that the temperature 
at different parts of L could be determined. The bulb A was placed at 
a portion of the furnace which by previous trial had been found to have 
a nearly uniform temperature. 
The temperature at the ends of A was two degrees lower at 400° C., 
and five degrees lower at 900° C., than at the middle. 
For the calculation of the vapour density the temperature of the 
vapour in A was taken to be the mean of the temperature at the middle 
and of the temperature at either end. 
The thermocouple used had been calibrated at the Physik. Techn. 
Reichsanstalt in Berlin, up to 1400° C. The E.M.F. was measured with 
a Siemens and Halske Millivoltmeter which was tested by direct com- 
parison with a standard instrument. 
The cold joints were immersed in water of approximately 20° C. 
contained in vacuum vessels. Their temperature was measured at each 
determination. 
The arrangement was also tested by direct comparison with a Zincke 
