Discharges in Insulators. 439 



traced, and the potential itself to be accurately determined at 

 the moment of the discharge. I hoped in this way to obtain, 

 with plates of different substances and especially with plates 

 cut in different directions out of the same crystal, a charac- 

 teristic difference of potential for each substance and for each 

 direction ; but hitherto all my endeavours have been fruitless. 

 I found it impossible, with one and the same plate, to obtain 

 satisfactorily accordant values from the different experiments 

 of one and the same series. The cause of this irregularity is 

 doubtless to be sought in an unavoidable difference in the dis- 

 position of the electricity on the points and the plate. The 

 difference of potential necessary for a spark-discharge is essen- 

 tially dependent on this disposition, which, in the method of 

 experiment chosen, changes before the spark passes, in con- 

 sequence of a less or greater conductivity of the plate and its 

 surface, as well as in consequence of electricity added by con- 

 vection from the point in an irregular and uncontrollable 

 manner. Perhaps, if experiments were made with much larger 

 plates and with very obtusely pointed electrodes, more favour- 

 able results might be obtained. 



The experiments which I made with liquids are, notwith- 

 standing their number, still too imperfect, and offer too few 

 general points of view, for their details to be communicated. 



As is well known, electrical discharges in gases have often 

 been the subject of investigation ; both the spark-discharge 

 with greater and less pressures and also the slow discharge 

 known by the name of dissipation have been repeatedly ex- 

 amined. From these experiments no simple relation can be 

 with certainty deduced between any constant of the different 

 gases and the difference of potential corresponding to each 

 gas, necessary for a discharge, or the amount of electricity 

 discharged. Yet it would be hazardous to conclude, on the 

 ground of those experiments, that such a relation does not 

 exist. For, in the first place, with spark-discharges it is 

 always to be feared that the decomposition which doubtless 

 takes place in some gases, as well as the considerable alteration 

 of temperature in the path of the spark, may possibly conceal 

 any such relation ; and, secondly, some hitherto unpublished 

 experiments made by M. Warburg have shown that in gases 

 dissipation cannot with certainty be demonstrated ; while the 

 loss of electricity by conductors which are insulated in gases, 

 observed by Coulomb, Riess, Warburg, and others, is very 

 probably brought about only by the insulating supports and 

 by particles of dust*. 



I therefore, after numerous preliminary experiments and 

 * See Boltzraann, Pogg. Ann. vol. civ. p 415. 



