64 DRS. J. A. HARK K I. 1 AND P. CHAPPUIS ON A 



repeatedly to temperatures varying from 200 to 250, showed that the volume of the 

 gas regularly diminished. 



It therefore seems evident from these experiments that the walls of " verre dur " 

 absorb a minute quantity of hydrogen. 



It appears probable that this absorption is due to the reduction of sulphates 

 contained in the glass. The employment of lead-glass as the material for the 

 reservoir instead of " verre dur " would probably give rise to still more serious effects 

 on account of the reduction of the salts of lead. 



To avoid in the measurement of temperature the uncertainties caused by the 

 variations of the gaseous mass, of which we have just spoken, and which might affect 

 not only its quantity but its composition, we have substituted nitrogen for hydrogen. 

 The nitrogen scale certainly diverges a few thousandths of a degree from the hydrogen 

 scale in the interval to 100. Its departure from the normal scale at high tem- 

 peratures is likely to be small and can always be corrected subsequently, when the 

 necessary data have been collected. 



The initial pressures of the nitrogen gas thermometer show no diminution, but 

 rather a slight increase, which is explained by the contraction of the glass due to the 

 annealing. 



XXV. COMPARISONS OF THE PLATINUM THERMOMETERS K.8 AND K.J) WITH THE 



MERCURY STANDARDS. 



The direct comparison of the platinum thermometers with the large normal 

 hydrogen thermometer between and 100 would have necessitated such an enormous 

 amount of work, without offering any special advantage, that we decided not to 

 employ this instrument, but to take instead the four primary mercury standards of 

 the Bureau, Tonnelot thermometers Nos. 4428, 4429, 4430, and 4431, whose cor- 

 rections to the hydrogen-scale have been previously determined with all possible 

 precautions by one of us. An account of this work is given in vol. 6, ' Trav. et Mem. 

 du Bureau International.' 



The comparisons between these mercury standards and the platinum thermometers 

 were made in an apparatus constructed originally for the comparison of mercury 

 thermometers with each other, which was modified and considerably improved for the 

 purpose of this research. This apparatus is shown in fig. 11. 



It consists of two concentric, rectangular, copper troughs; the outer one, which is 

 protected by an oak case, having a capacity of about 70 litres. This trough commu- 

 nicates by a side tube with a small vertical copper vessel, well protected against 

 radiation, which can be heated by a large gas burner. A screw stirrer, worked by a 

 small motor, drives through the heater a rapid current of water, which is taken in 

 at the opposite end of the trough by a horizontal tube resting on the lx>ttom, and 

 circulates as shown by the arrows in the figure. 



