335 
OF THE FLAME IN THE EXPLOSION OF OASES. 
glass was packed gas-tight by indiariibber, as in fig. 36. The gases were ignited by 
the spark at a, and the flame travelled towards h. Fig. 35 shows the result of firing 
equal volumes of cyanogen and oxygen in this apparatus. The photograph does 
not show the spark or first part of the flame. 
The flame, starting from the right-hand lower 
corner, is seen to become more luminous, and to 
move faster as it approaches the middle of the 
tube. The detonation-wave is then set up, its 
initiation being marked by a luminous-wave 
thrown backwards from the point where the 
detonation begins. To the end of the first glass tube the detonation-wave then 
travels with uniform speed and luminosity. On emerging from the lead bend into 
the second glass tube the flame is seen to be less luminous, but in a short period 
detonation is again set up, with the accompaniment of the backward thrust, and with 
the formation of a dark patch of gas, which is also characteristic. On passing the 
second lead bend the luminosity and speed are again reduced, and the detonation- 
wave is not again determined until the flame has traversed half the third glass tube. 
It was now evident that what liad happened in our “ collision” photographs was this 
damping down of the detonation-wave as it passed througli the lead bend and entered 
the glass tube. It was, however, difficult to imagine that the curvature of the tube 
had affected the rate, for Berthelot had determined the rate of explosion of electro¬ 
lytic gas in a leaden pipe with many sharp bends, and I had found the same velocity 
in a straight pipe and in one coiled I'ound a small drum. An experiment made Avith 
a glass tube bent to the same curvature showed no retardation (fig. 37). Berthelot 
had also found the velocity of explosion unaltered in a rubber tube Avith many bends; 
and when we proceeded to substitute a stout rubber tube for a lead bend (fig. 38, 
Plate 14), we found the AvaAm traversed it Avithout retardation. Therefore it was also 
equally hard to imagine that the use of a rubber tube to make a gas-tight joint 
betAveen the glass and the lead could affect the explosion, and yet this turned out to 
be the origin of the disturbance. 
We had connecting-pieces made of different metals and of varying curves and 
angles, but after many trials Ave found that the ordy thing Avhich mattered Avas the 
rigidity Avith Avhich the glass and metal AA^ere connected together. Any packing 
(such as rubber) Avhich gave to a shock caused a reta.rdation; Avhen the glass AA^as 
firmly cemented to the metal no retardation occurred. Fig. 39 shoAvs the explosion 
passing through two glass tubes rigidly connected to a copper junction, which Avas 
equally without retarding action whether curved or bent at a sharp angle. This 
retardation was not confined to cuiumd junctions. When a straight piece of lead 
pipe was fastened by rubber to two glass tubes, the wave of detonation was damped 
down on traA'ersing the junction; Avhen the metal AA^as cemented to the glass no 
retardation occurred. 
