730 Mr. Gr. Stead on the Separation of 



General Remarks. — In my previous paper (loo. cit.)\ have 

 shown that, when a discharge is passed through hydrogen 

 chloride, the chlorine spectrum appears at the anode, whereas 

 in chloroform, chlorobenzene, stannic chloride, &c, it is 

 found chiefly at the cathode. I have also pointed out that 

 chemical considerations indicate very conclusively that the 

 chlorine in hydrogen chloride carries a negative charge, but 

 that in the other substances referred to it carries a positive 

 charge. The results of the work described in the present 

 paper show that the chlorides of ethylene and ethylidene, and 

 likewise phosgene gas, behave in a manner analogous to that 

 of chloroform &c. Thus it would seem very probable that 

 the sign of the charge carried by the atoms is of fundamental 

 importance in connexion with the separation of spectra in a 

 discharge-tube. 



At the same time it appears impossible to regard this as 

 the only cause of the differences in the spectra at the anode 

 and cathode, since such differences are often found in an 

 elementary gas. Thus in oxygen the negative pole shows a 

 band spectrum and the positive pole a line spectrum ; in 

 nitrogen the band spectra at the two electrodes are altogether 

 different, whilst in hydrogen I have found perfectly definite 

 differences. Professor Sir J. J. Thomson also showed a good 

 many years ago that the spectra of this gas at the two 

 electrodes are not identical. The anode favours the pro- 

 duction of the secondary spectrum of hydrogen, and is 

 particularly unfavourable to the blue line H/3 (4861) of the 

 ordinary spectrum. The four-line spectrum is well developed 

 at the cathode, the blue line being here especially bright. 

 Now we can scarcely regard the blue line as being due to 

 positively charged hydrogen atoms and the red line Ha (6563) 

 to negatively charged hydrogen atoms, because these lines 

 both belong to the same spectral series, and therefore must 

 be presumed to have the same origin. Hence the differences 

 in hydrogen can be hardly assigned to electrochemical 

 effects. 



Now it is known that the secondary hydrogen spectrum is 

 favoured by a weak discharge and the primary spectrum by a 

 strong discharge. Again, in oxygen the negative band 

 spectrum has been obtained away from the negative pole, 

 indeed in the absence of any poles at all. Thus Donaldson * 

 got it with the electrodeless ring-discharge at an intensity 

 intermediate between that required for the compound line 

 spectrum and that necessary to produce the elementary 

 line spectrum. It seems therefore that the intensity of the 

 * Vide supra, p. 724. 



