ox THE MEASUREMENT OF TE1\[PERATURE. 
139 
Ttl 
employed in these researches, the larger thermometer being only used to determine 
four points on the vapour-pressure curve for licpiid oxygen on the hydrogen scale as a 
check on the measurements made by means of the smaller instruments. It will Ije 
remembered that in the large instrument the mercury column and dead-space were 
enclosed in a water-jacket, and the pressure was 
measured Ijy means of a scale of sj^ecial construction, 
which formed the first surface of the water-jacket. 
In these measurements the bulb of the ther¬ 
mometer and about 20 millims. of the stem were 
immersed in a liquid, consisting mainly of oxygen, 
contained in a vacuum vessel holding about 450 cub. 
centims., which was enclosed in a larger vacuum 
vessel containing a little liquid air. Beside the ther¬ 
mometer was placed, as in fig. 1, a glass tube m, in 
which pure oxygen, obtained by heating potassium 
permanganate, could 1)e licpiefied. This tube com¬ 
municated with the ])ump, witli the a])paratus for 
U'eneratino- the o-wuen, and witli a manometer of the 
type shown in fig. 2. During an experiment a stream 
of liydrogen As^as passed through the liquid in the 
inner vacuum vessel, to prevent superheating, and to 
stir it thoroughly.' 
In calculating the results it was possible to con¬ 
sider the whole of the stem from the toji of the bulb 
to the level of the top of the vacuum vessel as at the 
temperature of liquid air. The portion not immersed 
in the liquid was only about 30 millims. long, and the 
error so introduced would be considerably less than 
0'01° C. The coefficient of expansion of the glass 
between the freezing-point of water and the temperature of liquid air was taken as 
0’0000218, a number which we obtained by actual experiment (p. 138). The 
thermometer was filled with pure dry liydrogen by the method described in the 
second appendix to Part III. of this paper. 
The freezing-point of water on the constant-volume hydrogen scale is taken as 
273'03, the reciprocal of the pressure coefficient 0‘00366255. 
Tlie formula employed in calculating the results is as follows :— 
Fig. 1. 
Fig. 2. 
273-03 ^ 273 + T, ^ •' 
V,,, 
•■^73 -f D, 
= P 
V(1 - a (273 - T)) 
V, 
273 9- T', ' 273 + TbJ 
where P^ is the pressure on the gas when the bulb is surrounded with melting ice ; 
P is the pressure on the gas when the buHi is at the temperature to be measured ; 
T 2 
