326 Prof. J. J. Thomson on the Discharge 



sparks began in the dry did not seem to have any relation 

 with the corresponding potential for the wet gas. 



These experiments seem to show that hydrogen very 

 carefully dried is able to sustain a much greater potential- 

 difference than when containing a small amount of moisture, 

 but that the gas when under this exceptional potential-dif- 

 ference is apparently in an unstable state, as when a spark is 

 once started the potential difference at once sinks to about the 

 normal value in a gas containing traces of moisture, and this 

 potential-difference is sufficient to produce a second spark, if 

 this follow r s the first after only a short interval. 



It should be observed that the hydrogen has to be very 

 carefully dried in order to show these effects; it is not suf- 

 ficient, for example, merely to allow the gas to bubble through 

 sulphuric acid, gas so treated seems to differ little in its 

 electrical properties from undried gas. 



In order to see whether the effect produced by the presence 

 of water-vapour was due to the behaviour of the vapour 

 itself under electric strain, or to some interaction between the 

 vapour and the hydrogen, a bulb was prepared w T hich con- 

 tained nothing but water and its vapour. This bulb was 

 made by completely filling a bulb provided with electrodes 

 with distilled water, which was vigorously boiled for about 

 2^ hours until only a fraction of the water was left; this 

 bulb w r as sealed off whilst the water was vigorously boiling. 

 On making experiments similar to those described above, 

 the ratio of the difference of the potential required to produce 

 the first to that required to produce the second spark was 

 found to be at least as great as in the case of hydrogen. In 

 the water-bulb the pressure was only that due to the vapour 

 of water at about 18° C. We get the same difference in the 

 behaviour of the first and second sparks if we use a mixture 

 of gases instead of a single gas such as hydrogen. I find 

 that the effect is very marked in air that has been carefully 

 dried and filtered. 



These experiments show that the behaviour of a gas with 

 reference to the passage of a spark through it is analogous 

 to that of a vapour condensing to a liquid, to the freezing of 

 a liquid, or to the deposition of crystals from a saturated 

 solution. In all these cases, when no foreign substances are 

 present the temperature can be lowered far below the boiling- 

 point, the freezing-point, or the temperature at which depo- 

 sition takes place respectively without the corresponding 

 change of state taking place. When, however, foreign sub- 

 stances w r hich can act as nuclei are present, the change of 

 state takes place at a definite temperature. In the case of 



