Experiments on Positive Rays. 243 



pure oxygen ; i£ it is not ozone, the alternative seems to 

 be H ? O s . 



With mixtures o£ hydrogen and oxygen we get the water 

 line, m/e = 18, and sometimes a faint line for which m/e is 

 about 20 ; the compound H 4 would satisfy this condition. 



With pure oxygen, or with a large excess of oxygen, the 

 mercury lines disappear, owing I suppose to the combination 

 of the oxygen and the mercury vapour. With new aluminium 

 cathodes I have found that, whatever be the gas in the tube, 

 the mercury lines are either absent or very faint ; they attain, 

 however, considerable strength after the tube has been used 

 for some time and the electrodes have got saturated with 

 mercury vapour. 



When the discharge passed through CS 2 , I found the line 

 m/tf = 44 very strong ; this, I think, is due to the new com- 

 pound CS discovered by Sir James Dewar and Mr. H. 0. 

 Jones. Special precautions have to be taken in this case, as 

 C0 2 gives the same line ; oxygen was carefully excluded, 

 without affecting the strength of the line, and in this case 

 the line due to CO was exceedingly faint. This is a very 

 searching test, as when C0 2 is in the tube, the line due to 

 CO is always very strong, generally stronger than the one 

 due to C0 2 itself. 



The ammonia line was found when mixtures of nitrogen 

 and hydrogen were in the discharge-tube. 



It is quite possible for an element to be in the tube without 

 producing its line on the photographic plate. Thus I made 

 a number of experiments with nickel carbonyl in the dis- 

 charge-tube with the object of determining the atomic weight 

 of nickel by this method, but I was never able to detect a 

 line corresponding to nickel *. 



Again, when zinc ethyl is in the tube, I have never been 

 able to detect the zinc line. 



Indeed, with the exception of mercury vapour, the lines 

 due to heavy molecules or atoms are not so conspicuous as 

 we might expect. I attribute this to the peculiarities of the 

 photographic plates, which seem much more sensitive to 

 rapidly moving light atoms than to heavy ones with the same 

 amount of energy but a smaller velocity. 



As the photographic method, although excellent for de- 

 tecting the existence of different substances in the discharge- 

 tube, is not suitable for quantitative measurements of their 



* Since this paper was communicated the nickel lines Lave been 

 detected when a stream of nickel carbonyl ions keep running through 

 the tube and the discharge is kept on for a short time with long intervals 

 between the separate discharges. 



R2 



