45^ 



NA TURE 



[September 5, 1895 



two electrodes, then introducing by the side tube a very small 

 quantity of gas into the main tube, and again observing the 

 spectra at the two electrodes. 



A tul>e was filled with hydrogen and showed no trace of the 

 chlorine spectra : a ver)- small quantity of chlorine was then 

 let in through the side tut>e (in performing this cx[5eriment it is 

 necessary to be careful that only a very small quantity of chlorine 

 is introduced). After the discharge had been running through 

 the tube for a short lime, the chlorine specinmi was found to be 

 bright at the positive electrode, though no trace of it could be 

 detected at the negative. When the discharge was kept on for 

 some lime, the chlorine siiectrum, though still visible at the 

 positive electrode, got fainter ; it did not appear at all al the 

 negative. If a considerable quantity of chlorine was introduced 

 through the side tube, the chlorine spectrum was visible at both 

 electrodes, though it was brighter at the positive than at the 

 negative. 



When the induction coil was reversed, so that what was 

 before the positive electrode became the negative, the first 

 effect observed was that the chlorine spectrum flashed out with 

 great brilliancy at the old positive electrode, and was much 

 brighter than at any previous period. This, however, only lasted 

 for a second or two ; the chlorine spectrum rapidly faded away 

 and for a time was not visible at either electrode. Soon, how- 

 ever, the chlorine spectrum appeared at the new positive 

 electrode, having thus been transferred from one end of the lube 

 to the other. 



On again reversing the coil the same phenomenon was 

 repeated. There is ap]xirently no limit to the number of 

 times this effect may be obtained ; at any rate, I have driven 

 the chlorine from one end of a lube to the other 14 limes in 

 succession by reversing the coil. The chlorine is always driven 

 to the positive electrode, showing that the chlorine ion carries a 

 charge of negative electricity. The same effect was obtained 

 when a little vapour of bromine was introduced into the tube 

 instea<l of chlorine. When, however, the capillar)- tube was 

 filled with chlorine instead of hydrogen, and a little vapour of 

 bromine lei into the tube, the bromine went lo the lugath'e 

 eU,lroJc instead of to the |X)sitive, as it did when introduced 

 into the hydrogen tube. These experiments suggest that the two 

 gases in the tube combine, and thai the compound gas so formed 

 is split up into ions which travel along the lube ; that bromine 

 when in combination with hydrogen is the negative ion, and 

 therefore travels to the positive electrode ; when, however, it is 

 in combination with chhjrine the bromine is the positive ion and 

 travels to the negative electrode. 



.■\nother experiment tric<l was lo let a little vapour of sodium 

 into the middle of a capillary lube filled with air at a low 

 pressure. To prevent the sodium vapour condensing on the 

 walls of the tube, the whole tid)e was placed on a sand bath and 

 the lem|)eriiture raisc<l so Ijigh thai no conilcnsalion took place. 

 After the ilischarge had nm through the tube for about two 

 hours the sand was removed from the tube, and the movement 

 of the SfMliuin va|)our lo the mnative electrode was very apparent 

 even without using a spectroscope, as there was a great |xitch of 

 yellow light near the negative electrode and none in any other 

 |>art of the lulx:. 



.Vnolher ex|H;rimenl was lo intrtxluce a small (|uanlily of 



hydrogen into a lulie filled with air at a low pressure : the 



' '" _'en made its way lo the rici^ath't electrode. This experi- 



.1 ■.oniewhat troublesome one, as it is exceedingly difli- 



^;et these very fine capillary tubes sodry th.-il the spectrum 



oi the ilischarge does not show the hydrogen lines even before 



the hyilrogcn is intrrKluce<l into the middle of the lulic ; indeed, 



I never succeeded in getting rid of the hydrogen lines at the 



very lowest pressures. By healing the tube and allowing dry 



air to niti through it for a long tiniv, however, I got the tube so 



firy thai it tliil not shttw the hytirogen lines al a pressure fjuile 



1,,^.. ,.. h to allow the discharge to pass freely through it. 



lie was in this state and hydrogen was let into the 



lie lulic, the hydrogen s|ieclrum ap|X'ared al the 



' ■. but noi at the iKisitive. 



' of hydrogen al the negative electrode when 



' ilie with other gases has liecn descril>cd 



I \\ interesting |)a|>er in the J'/iilosopliual 



■ '■ 200. 



I ■ riments suggest, I think, that ihis 



. A and B, by the discharge is due to 



11 liy the discharge of a chemical com|»und 



H, in which the A atoms have a charge of 



electricity of one sign, the B atoms a charge of electricity of the 

 opposite sign ; these charged atoms under the influence of the 

 electromotive force in the lube travel in opposite directions, 

 Further, it follows from the experiment with the bromine vapour 

 in an atmosphere of chlorine that the sign of the electrical 

 charge on an atom of the same substance is not invariable, but 

 depends on the substance with which this atom is in combination. 

 We shall find numerous other instances of this change in the 

 sign of the charge on an atom in experiments described in a later 

 part of this paper. 



Polarisation of the Electrodes. — This in the electrolysis of 

 liquids is due lo the accumulation at the electrodes of ions which 

 have ceased to act as carriers of electricity. We have, I think, 

 distinct evidence of a similar accumulation in the electrolysis of 

 gases. Kor, as has been already described, after the discharge 

 has been running for some lime in one direction, giving the 

 spectrum of some gas al one of the terminals, the spectrum of 

 the gas at that terminal is momentarily brightened to a very 

 great extent by suddenly reversing the direction of the discharge. 

 After the current has been flowing for some time in one direction 

 through, say, CI in an atmosphere of H, the spectrum of the 

 chlorine, though still visible at the positive electrode, gets (\iinl, 

 the chlorine apparently to a great extent ceasing to carry the 

 discharge ; when, however, the current is reversed, the atoms of 

 chlorine can move freely, as they are not obstructed by the elec- 

 trode, so that immediately after the reversal of the current there 

 is probably more of the discharge carried by the chlorine than 

 at any other time, and the chlorine spectrum is consequently 

 brightest. 



Disiharge through a Compound Gas. — The separation of the 

 ions by the discharge can be readily observed in a tube of the 

 kind shown in Kig. 2. 



It diflers from an ordinary discharge tube merely in having a 

 flat metal plate, A is, fastened across the lube. When the dis- 

 charge passes through the lube, one side of the plate .ids as a 





NO. 1349, VOL. 52] 



Fig. 



positive, the other as a negative, electrode. The tube is mounted 

 on a stand, which the observer at the spectroscope can move by 

 means of a lever so as to bring one side or other of the 

 |)late opposite the slit of the sjiectroscopc ; a very slight move- 

 ment of the lever is suflicient lo do this, so that the spectra 

 al the two sides of the plate can readily be compared. I found 

 that the results were more satisfactory when the current was 

 kept flowing through the lube in one direction and the tulie 

 moved so as to bring the spectra at the two electrodes into the 

 field of view than when the tulie was keiit fixed in one position 

 and the current reversed. The latter method, however, sullices 

 to show the separation of the ions in many cases, and it has the 

 advantage of not reipiiring a [ilate across the tube ; all that is 

 necessary is to use for one of the terminals a disc whose plane is 

 parallel to the slit of the spectroscope. 



If the plate A 11 is thin, it is necessary to fuse it into the glass 

 tube all the way round ; otherwise, when the pressure is low, 

 the discharge, instead of crossing the plate, goes througli any 

 little crevices there may \k between the plate and tlie tube. 

 The easiest way of making the tube is to use a plale alioul 

 0'5 cm. thick, cut from an aluminium cylinder which tightly fits 

 the lube ; with n plale of this thickness the narrow spaces 

 between the tube and the plale are so long that the discharge 

 goes through the plale rather llian through tlie crevices. 



The lube was filled with the gas to be observed and the 

 spectra at the two sides of the plale compared. These spectra 

 were in many cases found to differ in a very remarkable way ; it 

 was, however, cudy in exceptional cases that a line which was 

 bright al one side of the plale was absolutely invisible on the 

 other. The method used was to lake two sets of lines, say A 

 and B, as close together in the spectrum as possible, and com- 

 pare the brightness of these sets of lines on the two sides of the 

 plate ; if it was founil thai the A lines were brighter on tlie 

 positive side of the plale than on the negative, while on the 

 other hand the B lines were brighter on the negative side of the 

 plate than on the |>osilivc, then it was inferred that electrolytic 



