OXY-ETHER AND HYDRO-CARBON LIGHTS 95 



led to the ' short circuiting ' of the oxygen-stream, and thereby 

 caused that predominance of oxygen which had somehow, it waa 

 evident, filled the gas-space over the tank with explosive mix- 

 ture, instead of merely combustible vapour. At last the tank 

 was ripped entirely open, and examined by an engineer very 

 closely, when all was explained. In the soldering of the parti- 

 tion which divided the safety chamber over the tank (as last de- 

 scribed) into two, as an entering and exit chamber, a very small 

 leak was discovered, with traces of corrosion. It is probable 

 that this at first was absolutely imperceptible, but had 

 gradually enlarged during the summer months ; and the re- 

 sult was that a certain amount of oxygen passed directly to 

 the exit chamber, without passing over the ether at all. 



Fro. 55. Ether Tank, side view 



The remedy against this danger, as was immediately 

 pointed out by Mr. Hardwich, was to entirely separate the 

 two chambers as in fig. 55. With this construction, another 

 such accident would be impossible. With regard to the 

 soldering of the partitions in the tank, any single leak between 

 two compartments would not have very much consequence ; 

 and it is the sides and top of the partitions which would need 

 special care. A little leakage at the bottom of the tank, ^hich 

 should be the last side to put on, would have no result, since 

 there would always be ether above that level, unless it were 

 actually exhausted, which never ought to occur. 



52. Porous Generators. Mr. Ives constructs his gene- 

 rators differently, filling a metal tube of two inches or more 



