the Luminosity of Gases. 249 



assumed by Pringsheim (loc. cit.), that we are justified in 

 speaking of the temperature of a flame as the temperature 

 indicated by thermometric instruments when they are intro- 

 duced into the flame. This seems to me to be fundamentally 

 erroneous. Such a recorded temperature is no doubt the 

 effective temperature of the flame, but it is merely an average 

 temperature. 



Every chemical student knows that the thinner the platinum 

 wire he uses the hotter it becomes when introduced into a 

 Bunsen flame, and it matters not whether the measurements 

 of temperature depend on the fusibility of metals or salts, on 

 specific heat, expansion, thermoelectric junctions, or electrical 

 resistance, they are all open to the obvious objection that they 

 only give the mean temperature of a considerable region of 

 the flame uncorrected for conduction losses. What are we to 

 say of the temperature of a candle-flame which will not melt 

 a small bead of common salt but which will melt a platinum 

 wire if pure and of a certain fineness ? 



We may best deal with the matter by considering an ideal 

 flame. Let us suppose that we have hydrogen issuing from 

 a vertical cylindrical tube into oxygen gas, and let us consider 

 a horizontal slice of the gas one molecule thick moving up- 

 wards in the tube. 



For the sake of simplicity we may neglect the molecular 

 motion of the gas, and we will also suppose the gas to inflame 

 spontaneously on meeting with the requisite oxygen. The 

 slice of gas as it ascends the tube is retarded by friction at 

 the edges and will issue with a conical form from the tube. 

 Here it inflames and gives us a small conical flame. This 

 flame would consist of a conical sheet of hydrogen, one mole- 

 cule thick,, combining with oxygen. If we are to measure 

 the temperature at the locus of combination we must have a 

 thermometric instrument small compared with the thickness 

 of the sheet of combustion. If the instrument does not fulfil 

 this condition, it will protrude into the unburnt gas within the 

 cone and the cooling gases outside, and the temperature indi- 

 cated will be the average of the whole region in which it is 

 immersed. 



It may be objected that this is only an ideal flame, and that 

 the conical sheath of flame actually obtained when hydrogen 

 burns in oxygen has a very sensible thickness. This is no 

 doubt true ; but it must be at once pointed out that if the 

 sheath of burning hydrogen has a sensible thickness, that is to 

 say if the hydrogen is burning on the outside surface, the 

 inside surface, and in the intervening thickness of the sheath, 

 hydrogen must all the while be passing through the sheath 



