440 Mr. G. J. Stoney on the Penetration of 



nus, and others, have been attributed to copious conduction of 

 heat through hydrogen. May we not with more probability 

 refer them to the very remarkable power which hydrogen 

 possesses of allowing heat to leak aw^ay by penetration? 

 From the dynamical theory of gases, it seems improbable that 

 any gas can possess true conducting-power in a high degree ; 

 and every observer must have been struck by the violence of the 

 first chilling effect in some of these experiments, and the 

 unflagging energy with which it is maintained, a promptness 

 and persistence characteristic of penetration, but quite unlike 

 the moderate initial effect and diminished subsequent progress 

 which we should expect from conduction. 



19. One of the most striking of these experiments is that 

 by Sir William Grove, which exhibits the cooling effect of 

 hydrogen on a wire rendered incandescent by the passage of 

 an electric current ; and according to the hypothesis here 

 presented it ought to be possible to repeat this experiment in 

 ordinary atmospheric air by bringing the incandescent wire 

 sufficiently close to a cool object. This very instructive ex- 

 periment was proposed by my son. Master Gerald Stoney, and 

 has been successfully performed by him. He first passed the 

 wire through a glass tube drawn out sufficiently thin. The 

 effect could then be seen ; but it was evanescent, because the 

 glass became rapidly heated. No doubt, if the tube had been 

 surrounded by a water jacket, the experiment might have been 

 made in this way satisfactorily. But he made it in an equally 

 permanent form, and with greater ease, by simply bringing the 

 incandescent wire close to a tin can containing water, to which 

 the heat leaked away abundantly from the wire when the in- 

 tervening stratum of air was sufficiently thin. The effect is 

 best seen when the wire is of a dull red, on account of the 

 ease with which the eye detects the difference between dull 

 red and darkness. It then becomes conspicuous when the 

 interval is a millimetre, and can be perceived when the in- 

 terval is considerably more. In this experiment the can was 

 at a slightly higher temperature than the room. On this 

 account, and because a small part of the radiated heat was 

 reflected back on the wire by the tin, the loss of heat by 

 radiation was less than when the can was away; moreover 

 the convection current was enfeebled by being both cooled and 

 obstructed by the neighbouring obstacle. Hence the true loss 

 of heat by penetration must have been in excess of that which 

 manifested itself. 



20. Another phenomenon which admits of explanation by 

 the theory developed here and in my former papers, is one 

 which is said to have caused the bursting of steam-boilers, 



