150 PROFESSOR H. B. DIXON" OX THE RATE OF EXPLOSION IN OASES. 
the interval during which the gas remains bright enough to effect a sensitive 
isochromatic plate cannot be more than one five-thousandth of a second. 
The problem of measuring the pressure in a thin layer of gas, lasting for so small 
a fraction of a second, is one of the greatest difficuitj. I believe the method tried by 
Mallard and Le Chatelier for explosions of carbon lasulphide and oxygen is 
probably the best that has yet been devised. This depends on the principle that if a 
pressure is j^roduced in a glass vessel greater than the glass can stand, the vessel will 
be broken, although the pressure may endure for a very small interval of time. If the 
vessel is broken by the explosion, T believe that there must have been a jDressure 
produced in the layer of gas next the glass at least equal to that which would break 
the vessel when gradually applied. By compressing gas in the vessel by means of a 
pump, it can be shown that the vessel is able to withstand a certain pressure. The 
vessel can then be connected with a long tube, the whole filled with an explosive 
mixture, and the explosion-wave set up in the tube. If the vessel is fractured, 
a pressure was exerted on its walls greater than the pressure applied by the pump. 
To test this method, I have carried out a few preliminary experiments. Out of a 
piece of strong combustion tubing I made twelve short tubes, carefully sealed at one 
end, and slightly opened at the other. These tubes were then firmly attached to a 
metal piece by a collar and nut v/orking a,gainst india-rubber rings—care being taken 
that the glass did not come anywhere into contact with the metal, but only with the 
rubber rings. The metal piece could be attached either to the pump or to a leaden 
firing tube. The twelve tubes were first tested with compressed air, which was 
slowly forced into the tube while an observer read the manometer. One tube (it was 
rather unevenly sealed) broke down between nineteen and twenty atmospheres. A 
second broke at twenty-three atmospheres. The other ten stood a pressure of 
twenty-five atmospheres for two minutes. These tubes were then in turn attached to 
one end of a firing-tube about twenty feet long. On filling the whole with a mixture 
of carbonic oxide and oxygen, and passing a spark through the gases at the other end 
of the tube, the explosion-wave was propagated through the leaden pipe into the glass 
vessel. All ten tubes were blown into small fragments by the explosion. Four stronger 
green glass tubes were tested by compressed air at fifty atmospheres. All these tubes 
stood the explosion of carbonic oxide and oxygen, and hydrogen and oxygen. Two of 
these green tubes were blown to pieces by the explosion of equal volumes of cyanogen 
and oxygen, a third was fractured by compressed air at seventy-eight atmospheres. 
These experiments confirm Le Chatelier’s results with the Deprez indicator, showing 
that very high pressures are produced for the moment in the explosion-wave ; they do 
not, of course, measure the pressures, they only indicate a lower limit. 
The quiet burning of carbonic oxide and oxygen, when this mixture is lighted at 
the open end at a,.short tube, such as a test-tube, is in striking contrast to the violence 
of the explosion when the “ wave ” is set up in the same gases. As a lecture experi¬ 
ment a test-tube fidl of the mixture may be ignited by a taper, wdien the quiet 
