172 PHENOMENA ACCOMPANYING 



uometcr the effect of u continuous current. The passage of the discharges may 

 even be interrupted for many minutes without the loss of the capacity which 

 the gas had acquired of immediately transmitting them. To lose it completely 

 we must wait a long time, or renew the gas, and consequently again rarefy it. 

 An equally important fact to be noticed is, that the discharges once transmitted, 

 there may be gradually introduced, while they arc passing, a new quantity of 

 the same gas, which amounts to an augmentation of the density, without their 

 ceasing to pass; the pressure may thus be carried to almost double what it was 

 at the beginning. The direction of the discharges has no influence on this 

 train of phenomena; for, the discharge having once effected a passage, its 

 direction may be changed at will, without a cessation of immediate transmission. 

 This result, which 1 have had occasion to verify in many and very different 

 cases, would seem to show that the gaseous matter opposes a certain inertia to 

 the establishment of that particular disposition Avhicli the transmission of elec- 

 tricity requires, and which determines the tension which precedes that trans- 

 mission ; but that this disposition once established, it subsists long after the 

 passage of electricity has ceased, provided no disturbance intervene in the state 

 of the gas. It had long been supposed, particularly in the theory of Grotthus 

 on electrolytic decompositions, that something analogous occurred in the trans- 

 mission of voltaic currents through liquids ; it had thence been inferred that 

 the tension of the poles of the piles induced in the liquid in which these poles 

 were plunged, a polarization, which preceded the passages of the current. Only 

 these two effects succeeded one another in a time so short as to be inappre- 

 ciable, while with gases they would be found to be separated by an interval 

 of more or less duration, but always appreciable. 



I shall restrict myself here to certain numerical results, — results of but little 

 importance indeed, since it is impossible to deduce from them a law, in view of 

 the numerous causes which occasion them to vary. They serve only to show 

 the accuracy of the general principle which I have just indicated. We may 

 also infer from them the great superiority of hydrogen over nitrogen and atmo- 

 spheric air, as regards its conducting power — a fact already noticed by several 

 experimenters. 



In a tube 5 centimetres in diameter and 16 in length, the discharge, when 

 the tube was filled with atmospheric air, only began to pass when the pressure 

 was reduced to 20 millimetres ; with nitrogen it passed under a pressure of 24 

 millimetres, and with hydrogen under that of 36 millimetres. It is true 

 that subsequently, under the same conditions of intensity, and still with 

 hydrogen, the discharge passed at pressures of 42, and even 48 millimetres ; 

 when rendered still stronger, it has been transmitted under a pressure of 

 even 72 millimetres. With a tube having a like diameter of 5 centimetres, 

 but only one metre in length, the same discharge only began to pass through 

 nitrogen under a pressure of from 4 to 5 millimeters ; with hydrogen it passed 

 only under a pressure of from 12 to 13'""\ Afterwards, when stronger, and 

 again with hydrogen, it passed under a pressure of 18, and even 20°"'". When 

 the discharge begins to be transmitted, it exhibits itself in very minute jets 

 or streams, more or less intermitted ; afterwards these streams combine to form 

 a larger and more continuous one. In a jar iilled with hydrogen the discharge 

 passed from an isolated central ball to a ring 12 centimetres in diameter, 

 making the distance of the transmission but 6 centimetres in a space of hydro- 

 gen wliich might be called unlimited. In this case it was transmitted under a 

 pressure of 128""" in the form of streams more or less intermittent and undu- 

 lating, darting from the central ball to all points of the ring indifferently. At 

 90'""' the discharge gave rise to a continuous stream, susceptible of being in- 

 fluenced by magnetism. 



We see, by the instances just cited, that the pressure under which, fijr an)» 

 given gas, the discharge can pass varies with so many circumstances, that its 



