ON THE MEASUREMENT OF TEMPERATURE. 
137 
Hydrogen Thermometer. 
Pressure. 
Temperature. 
Boiling-point; calculated. 
° C. 
760-6 
- 182-20 
- 182-21 
764-4 
-182-67 
-182-72 
759-5 
-181-62 
- 181-52 
The coefficient of expansion for hydrogen is taken as 0’0036625. Taking the same 
coefficient for oxygen, a proceeding which does not appear to be justified, he oljtained 
the following results :— 
Oxygen Thermometer. 
Pressure. 
Temperature. 
Boiling-point; calculated. 
°C. 
° C. 
772-5 
- 182-29 
-182-43 
756-0 
-183-46 
- 183-42 
753-5 
- 182-95 
-182-90 
Baly (‘Phil. Mag.,’ 1900, vol. 49, p. 517) measured the vaj:)Our pressures of 
oxygen on the constant-j^ressure hydrogen scale, employing a Callendar thermoiiieter. 
The results appear to be afiected by a constant error; they will be discussed 
elsewhere. 
So far as the evidence contained in the original })apers is concerned, it appears to 
us that greatest weight is to be attached to the work of Holborn and Wien. It is 
true that the results of Wroblewski, Olszewski, and Estreicher show a close 
agreement, ljut none of these experimentalists furnish sufficient data to make 
criticism possible. 
In all the experiments we have referred to, the boiling-point or vapour pressure 
of oxygen has been measured by immersing the thermometer in the li(piid and 
ineasuring tbe pressure at which it was boiling. Estreicher and Olszewski 
measured tlie vajjour pressures at lower temperatures by connecting the vacuum 
vessel containing the licjuid with an exhaust pumj), and making tlie licpiid boil under 
reduced pressure. 
The difficulty of making accurate measurements by this method is twofold. In 
the first place, it is not easy to liquify a sufficient quantity of pure oxygen; and in 
the second place, the liquid oxygen tends to become superheated, and does not boil 
steadily. The impurity in the oxygen would most probably be air, -which would 
lower the boiling-point ; other impurities would, however, have the opposite effect. 
In every case it appears to have been assumed that the coefficient of expansion for 
glass remains constant over the whole range of temperature, 
VOL. cc. —A. T 
