to produce Discharges in Gases at Low Pressures, 497 



above the critical pressure, while in the single reading we have 

 below the critical pressure it is negative. It may be noted, 

 however, that the negative value of R is numerically smaller 

 instead of larger than the positive values, and i£ compared 

 with V 3 instead of with the average of Y x and V 3 , is only 

 three volts. Furthermore, this negative value was obtained 

 immediately after the set of readings in Table I., which gave 

 the unusually small value of 5 volts for R, and it is possible 

 that the conditions were in some way abnormal for both. 

 Evidence of a much more definite nature is found upon 

 examining the results obtained when the spark chamber is 

 not closed till it is time to stop the current. In this case, the 

 pressure has a chance to adjust itself all the time the current 

 is running, so that when the valves are finally closed and the 

 current stopped, the pressure and the kind of gas are the 

 same inside the spark-chamber as outside, even if some 

 hydrogen has been driven from the electrodes. Thus opening 

 the valves should make no difference, or at most a small dif- 

 ference, if the pressure has not become quite equalized, and 

 this small difference should bo of opposite sign above and 

 below the critical pressure. Table I, shows that the results 

 are the exact opposite of these. Instead of a zero value of R, 

 or small values which change sign at the critical pressure, we 

 have much larger values of R on the average than with the 

 spark-chamber closed all the time, and the sign of R does not 

 change at any point. From these considerations it appears 

 that we cannot explain the results by an increase in the 

 amount of hydrogen between the plates any better than by a 

 decrease. 



If an explanation is sought in the presence of some other 

 gas, it must be one whose sparking-potential curve lies above 

 that of hydrogen at all points, as R does not change sign at 

 any pressure, above or below the critical pressure. Since air 

 fulfils this condition, it was thought possible that there was a 

 slight leak in the black-wax joints of the spark-chamber, and 

 that the air thus admitted, though not enough to change the 

 pressure appreciably in the whole apparatus, might be suffi- 

 cient to have an effect in the small space of the spark- 

 chamber. Such an effect, however, would tie independent 

 of the presence of the current, and the fact that the sparking- 

 potential is sensibly the same whether the chamber is open 

 or closed, so long as the current is not run, but changes alter 

 the passage of the current, disposes of this possibility. 



Jf the effect were due to some gas of generally higher 

 sparking potential than hydrogen, such as air, driven from 

 the electrodes, one would naturally expect thai the effect 



