PRODUCED BY HOT PLATINUM IN DIFFERENT OASES. 55 



an earth-connected tube J in which it was held by an ebonite cylinder K. The whole 

 of this part of the apparatus was fixed in a wooden box covered with lead foil 

 connected to earth. 



In reducing the thermocouple readings to temperatures, the platinum temperature 

 was first calculated by making use of the reading corresponding to the melting-point 

 of potassium sulphate. This point was determined experimentally in the way already 

 descril>ed. The platinum temperatures were then reduced to centigrade by means of 

 the correction curve given by CALLENDAR.* The legitimacy of this process was 

 tested by making an independent determination of the melting-point of sodium 

 sulphate. The value found was 885 C., and is within 2 of that (883 C.) given by 

 HEYCOCK and NEVILLE for this constant. 



Sealed on to A' were a mercury manometer, a glass bulb of about 300 cub. centims. 

 capacity, and a glass tap. The last named was connected to the apparatus for 

 delivering and purifying the hydrogen which was prepared, as described previously, by 

 the action of pure zinc on hydrochloric acid. From experiments on the diffusion of 

 the hydrogen through the walls of the platinum tube it is l)elieved to have been 

 exceptionally pure. The tube A was also sealed on to a three-way tap so that the 

 hydrogen or air could be either sucked by means of a water pump or allowed to 

 bubble through water. These arrangements made it easy (1) to test if the hydrogen 

 was diffusing through the tube properly, (2) to replace the stream of hydrogen by air 

 and vici' versd, and (3) to change the pressure inside the apparatus which regulated 

 the rate of diffusion of the hydrogen through the walls of the tube. 



It is convenient to consider first the effect of the hydrogen on the negative 

 ionisation in air. Preliminary experiments showed that the current could not be 

 saturated by the voltages at the writer's disposal, so the current with 80 volts was 

 measured instead of the saturation current. This makes the absolute values of the 

 currents considerably smaller than those previously obtained, more especially as the 

 latter are probably greater than the normal on account of ionisation by collisions (see 

 p. 23). The first test was made at 1200 C. After the tube had been heated for a 

 long time with air both inside and outside, the current with 80 volts was found to 

 lie 21 x 10~ u ampere per square centimetre of surface. The tube was then allowed to 

 cool, the air replaced by hydrogen inside, and the leak again measured. With 

 hydrogen inside at a pressure of 115 millims. the leak under the same conditions was 

 26xlO~ u ampere per square centimetre, and with hydrogen at atmospheric pressure 

 24xlO~ 14 ampere per square centimetre. These numbers are all equal within the 

 proliable accuracy of the temperature regulation. 



Another experiment was made at 1380 C. The current with 80 volts with 

 hydrogen inside the apparatus at atmospheric pressure was found to be 37 x 10~* 

 ampere per square centimetre, and with the hydrogen at 65 millims. the current was 

 3'G x 10"" ampere per square centimetre. In the latter case the amount of hydrogen 



* 'Phil. Mag.' [5], vol. 48, p. 519. 



