1895.J On the Electrolysis of Gases. 251 



Influence of the Chemical Constitution of a Compound on the Sign 

 of the Charge of Electricity on one of its Constituent Atoms. In many 

 organic compounds an atom of thejelectro-positive element hydrogen 

 can be replaced by an atom of the electro-negative element chlorine 

 without altering the type of the compound. Thus, for example, we 

 can replace the four hydrogen atoms in CHi by chlorine atoms, getting 

 successively the compound CH 3 C1, CH 2 C1 2 , CHC1 3 , and CC1 4 . It 

 seemed of interest to investigate what was the sign of the change of 

 electricity on the chlorine atom in these compounds. The point is of 

 some historical interest, as the possibility of substituting an electro- 

 negative element in a compound for an electro-positive one was one 

 of the chief objections assigned against the electro-chemical theory of 

 Berzelius. 



When the vapour of chloroform, CHC1 S , was placed in the tube, it 

 was found that both the hydrogen and the chlorine lines were bright 

 on the negative side of the plate, while they were absent from the 

 positive side, and that any increase in the brightness of the hydrogen 

 lines was accompanied by an increase in the brightness of those due 

 to chlorine. The spectrum on the positive side of the plate was that 

 called the carbonic oxide spectrum ; when first the discharge passed 

 through the tube, the spectrum on the positive side was the 

 so-called candle spectrum, but this very rapidly changed to the 

 carbonic oxide spectrum. The appearance of the hydrogen and 

 chlorine spectra at the same side of the plate was also observed in 

 methylene chloride and in ethylene chloride. Even when all the 

 hydrogen in CH 4 was replaced by chlorine, as in carbon tetrachloride, 

 CC1 4 , the chlorine spectra still clung to the negative side of the plate. 

 To test the point still further, I tried the analogous compound 

 silicon tetrachloride, inserting a small jar in the circuit to brighten 

 the spectrum. The chlorine spectrum was again brightest at the 

 negative side of the plate, while the silicon spectrum was brightest at 

 the positive. This is a very favourable case for the application of 

 this method, as there are two silicon lines (wave-lengths 5058, 5043) 

 quite close to two chlorine ones (wave-lengths 5102, 5078), so that 

 their relative brightness can easily be compared. The experiment 

 with the silicon tetrachloride is more conclusive than those with the 

 carbon compounds, as with the latter the spectrum on the positive 

 side of the plate is a band spectrum, and since the potential gradient 

 when the discharge is passing is very much steeper on the negative 

 side of the plate than on the positive, the effects observed might be 

 supposed to be due to the circumstances on the negative side being 

 better adapted for the production of line spectra than those on the 

 positive. This explanation is not, however, applicable to the case 

 of silicon tetrachloride, where the spectra on both sides of the plato 

 are line spectra. 



