PRODUCED BY HOT PLATINUM IN DIFFERENT OASI - 



35 



low temperatures. The numbers given below show that at 1170 C. the leak is very 

 closelv iv|nvsi'iiti"l liv an expression of the form P/(/J+P), which would be the value 

 given by the theory if most of the gas outside the wire were dissociated. This is 

 likely enough to be the case at pressures of the order of 1 millim., and at higher 

 pressures the exact form of f(P) has very little influence on the result. No doubt a 

 better agreement could have been obtained by putting in the theoretical value for p> 

 viz., 



*k, but this involves introducing a new constant k into the expression 

 for the ionisation, a refinement which is not warranted by the exactitude of the 

 experimental results. It appeared to be impossible to test the theory more fully by 

 pushing the experiments to a higher order of accuracy, on account of the irregularities 

 which have already been described. 



The experimental values of the ionisation at 1170 C. and those calculated from the 

 formula aP/(/3 + P) are given in the following table, a was taken = 38 x 10~ w amplre, 

 and ft = 4 '8 millims. of mercury : 



The greatest value of the positive leak in oxygen at 1170C. works out at 

 17 x 10~ 8 ampere per square centimetre of platinum surface. The smaller amount of 

 variation with temperature exhibited by the constant ft is a surprising result. 

 At the two lower temperatures /? = Bi~'A~' in the notation previously employed, 

 and its constancy with temperature implies that the heat of formation of the 

 platinum-oxygen systems from atomic oxygen is nearly equal to the heat of 

 formation of an oxygen molecule from two atoms. At the higher temperature /? is 

 equal to B/A, and its value in conjunction with the preceding result implies that 

 B/A is small compared with 4 '8 at the lower temperatures. 



According to the present theory, the energy change associated with the action 

 which gives rise to a positive ion is given exactly by the temperature variation of the 

 maximum value of the leak and not by the temperature variation of the leak under 

 a pressure of I '17 millims. as was assumed on p. 25. The preceding results enable us 

 to correct for tins, hut as the effect of the correction is smaller than the probable 

 experimental error it has not been proceeded with. 



F 2 



