32

that the flame passed into the atmosphere/but did not communicate through the stop-cock, so as to inflame the mixture in thebladder: and in comparing the power oftubes of metal and those of glass, it ap-peared that the flame passed more readilythrough glass tubes of the same diameter ;and this phenomenon probably dependsupon the heat lost during the explosion incontact with so great a cooling surface,which brings the temperature of the firstportions exploded below that required forthe firing of the other portions. Metal isa better conductor of heat than glass : andit has been already shown that the fire-damprequires a very strong heat for its inflam-mation.”

I found, likewise, that inflammation couldnot be communicated to explosive mixturesin long narrow canals of metal, or in shortcanals of smaller diameter, or through finemetallic gauze of the kind used for sieves.

“ The consideration of these various facts,led me to adopt a form of a lamp, in whichthe flame, by being supplied with only alimited quantity of air, should produce sucha quantity of azote and carbonic acid, as to