through Rarefied Air. 319 



the inferior plate, that is to say, when the negative electrode had 

 a sufficiently large superficial area, of the many partial discharges, 

 which, with decreasing density, constitute the discharge of the 

 battery, a jmrt of the last, owing to their small tension, must 

 have passed as glimmering discharges, and thus have exercised 

 no heating effect. When, on the contrary, the du'ection of the 

 current was from the knob to the small platinum surface, the 

 negative electrode was limited in magnitude, and a part of the 

 glimmering must have been converted into discontinuous dis- 

 charges, and thus become effective in heating the circuit. Hence, 

 in accordance with experiment, when the discharge passes from 

 the sui'face to the knob, the heat in the remaining part of the 

 discharge circuit must be smaller than when the discharge has 

 an opposite direction. It is evident that the number of the 

 glimmering discharges which become effective by reversing the 

 current differs according to the form and distance of the elec- 

 trodes. Cases may present themselves where the difference in 

 the heating effect is inappreciable ; but wherever it does occur, 

 it must be in favour of the stated direction ; and this, without 

 exception, I have found experiment to corroborate. The expla- 

 nation above given is also supported by ocular evidence. If the 

 cylinder is placed in the shade, and observed during the time 

 that the discharge passes from the knob to the surface, a brilliant 

 light is noticed in the space between the electrodes, whilst the 

 knob and rod are but feebly luminous. It may here be remarked, 

 that this brilliant light between the electrodes consists of two 

 dazzling colours; at the negative electrode it is reddish-violet, 

 at the positive greenish-blue. When, on the contrary, the dis- 

 charge passes from the surface to the knob, a broad reddish 

 light is observable, not only between the electrodes, but through- 

 out the length of the glass cylinder, and stretching over the 

 knob and rod. In the last case, the explanation supposes that 

 a great part of the discharges are glimmering. 



In free air, when the discharge spark passes between a very 

 small surface and a knob, no difference between the heating 

 effects in the discharge circuit is observable when the direction 

 of the current is changed. We have yet to examine, therefore, 

 to what degree the rarefaction of the air must be carried in order 

 that this difference may be observable. The glass cylinder was 

 screwed on to the air-pump and kept in communication with the 

 barometer test, the foot and clamp of the cylinder were connected 

 with the discharge circuit of the battery, by means of long copper 

 wires, after the thermometer had also been introduced into it. 

 The knob in the cylinder stood at a distance of ^^ths of an inch 

 from the small surface. In the following table the heating effects 

 corresponding to the unit of charge are given, each of which was 



