154: Scientific Intelligence. 



than platinum at temperatures near the boiling point of hydro- 

 gen.* The difficulties of the investigation were considerable: 

 it had to be carried out at the time in the neighborhood of the 

 machinery producing the liquefied gases required in the investi- 

 gation, namely, oxygen, nitrogen, and hydrogen, so that the zero 

 of the delicate galvanometer employed did not remain quite con- 

 stant. To remedy this I inserted a rocking make-and-break in 

 order to get the readings of each observation at both ends of the 

 scale. In the process of removing one difficulty another pre- 

 sented itself, through the development of small but appreciable 

 thermo-electric currents in the rocker. Precautions had to be 

 taken against these and at all other metal junctions against 

 similar small thermo-electric currents, and it was even found 

 necessary to have a correction on account of the resistance box, 

 inserted in the circuit to bring large readings within the limits of 

 the scale. The galvanometer and resistance box were inserted in 

 the German-silver branch of the couple, the points of junction of 

 the copper leads with the German silver ends of the couple being 

 insulated and placed close together within a vacuum test-tube 

 packed with cotton wool to ensure equality of temperature. 



Preliminary experiments showed that the junctions altered after 

 having been subjected to the temperature of liquid hydrogen. 

 However, on re-soldering the junctions with hard silver solder 

 instead of soft solder, the thermocouple accurately repeated 

 observations at the temperature of liquid oxygen, after having 

 passed through a liquid hydrogen bath. From this it appears 

 that all such couples before calibration ought to be cooled sud- 

 denly in liquid air and then rapidly heated to the ordinary tem- 

 perature, a similar operation being repeated with liquid hydrogen. 

 If the couples return to their original state after such abrupt 

 changes of temperature, then they are in a fit state for calibration. 



Three series of observations were made to determine whether 

 the resistance of the junctions varied to a noticeable extent with 

 the temperature, namely, at the freezing point of water, at the 

 boiling point of oxygen, and at the boiling point of hydrogen. Six 

 very concurrent observations with varying resistances in the resis- 

 tance box were made between 0° C. and 15° C. These were reduced 

 by the method of least squares, and gave for the resistance of the 

 circuit 3*500 ohms. Five similar results between the melting 

 point of nitrogen and the boiling point of oxygen gave, by least 

 squares, 3*293 ohms. Only two observations were taken in liquid 

 hydrogen, which are therefore not entitled to the same weight as 

 those already given, but the resistance appeared again about 33 

 ohms. As the variation of the resistance of the circuit was so 

 slight, an attempt was made to reduce the results on the assump- 

 tion of constancy, but this was not satisfactory. However, on 

 treating the variation of the resistance of the circuit as linear 

 with the temperature, the data came into better agreement. 



*Bakerian Lecture, " The Nadir of Temperature and Allied Problems," 

 Proc. Roy. Soc, vol. lxviii, 1901. 



