PRODUCED BY HOT PLATINUM IN DIFFERENT CASKS. 



17 



400 

 Pressure 



Fig. 6. 



The independence of pressure exhibited by the leak at high pressures seemed at 

 first sight a very surprising result and led the author to enquire whether there might 

 not be something wrong with the temperature measurements at high pressures. As 

 has been explained, the criterion employed to obtain a constant temperature so far has 

 l>een a constant resistance of the wire. Now the resistance measures the average 

 temperature of the wire across its section, whereas what is required in these measure- 

 ments is a constant surface temperature, so that any change which alters the 

 temperature gradient from the centre of the wire to the circumference will alter the 

 surface temperature at constant resistance. Increasing the gas pressure facilitates 

 the flow of heat from the surface of the wire and must therefore increase the internal 

 radial temperature gradient. It is evident, then, that increasing the gas pressure 

 lowers the surface temperature when the resistance is kept constant. It might be 

 thought that this effect would be small in the thin wires used (O'Ol centim. diameter), 

 but tlif leak is a very rapidly varying function of the temperature, so a small 

 temperature error produces a big change in the leak. 



To eliminate this error, which only enters into the experiments on the pressure 

 variation, and then is only important at somewhat high pressures, a method was 

 devised by which the surface temperature was kept constant. A tube similar to that 

 shown in fig. 1 was constructed, exhausted, and sealed up. The wire in it was then 

 heated to a standard temperature by means of a constant current. A portion of this 

 filament was then compared with a similarly situated portion of that from which the leak 

 was being measured, and the heating current through the latter was adjusted until the 

 t \vo appeared to l>e of the same brightness. Both hot wire tubes were shut up in a 

 black-lined box, and by looking into this through a tube furnished with paper slits 

 the field of view could l>e limited to those portions of the filaments which it was 

 desired to compare. 



VOL. ccvu. A. D 



