294 Rev. P. J. Kirkby: Chemical Effects of the 



be determined by the final slope of the curve, if the curve 

 becomes ultimately straight. 



To construct such curves it is necessary to find the chemical 

 effects of passing the same current through hydrogen and 

 oxygen at the same pressure but with different distances 

 between the electrodes, and to find the corresponding poten- 

 tial-differences of the electrodes. As the discharge passes 

 the pressure steadily falls and the current changes slightly, 

 generally falling steadily. But if the variations in pressure 

 and current are fairly small, the results will apply with 

 sufficient accuracy to that pressure and that current which 

 are the mean values of the changing pressure and current. 



Several series of experiments were accordingly made, each 

 series with approximately the same mean pressure and current. 

 Moreover, the mean current was nearly the same throughout, 

 being in almost every case within 5 per cent, of the value 

 "00275 ampere, which was selected as the norm partly 

 because it was accurately read on the ammeter, partly because 

 it was easily maintained for a sufficient time in the hydrogen 

 and oxygen at the required low pressures, being neither too 

 great nor too small. 



The current, as has been said, slowly changed, but it was 

 stopped before it diverged more than about 5 per cent, from 

 its mean value. Thus the currents used in every one of the 

 experiments recorded in the tables below T lay always within 

 about 10 per cent, of '00275 ampere. It was an easy 

 matter to adjust the current by means of external resistances 

 except at very low pressures. But it is not so easy to secure 

 the desired mean pressure. 



, Let p inms. denote the pressure before the current was 

 started, and p x the pressure after it was stopped and the 

 water- vapour resulting from the discharge had been wholly 

 absorbed. Then the mean pressure 



Thus, to carry out an experiment with a given mean 

 pressure, p, p had to be estimated so that after the experi- 

 ment p should reach the desired value. This is a difficult 

 result to achieve, partly because it is hard to adjust pressures 

 by means of a mercury pump with sufficient delicacy, and 

 partly because the adjusting the current so that its mean 

 value should lie near enough to '00275 ampere frequently 

 involved a considerable fall of pressure. Nevertheless, the 

 mean pressure for almost all the experiments in each series 

 tabulated below was constant within 3 or 4 per cent. 



The fall of pressure, Ap, due to the absorption of the 



