and Heat-conduction in rarefied Gases. 59 



plane solid sides having the constant temperatures and t, we 

 see that, if with decreasing density the mean length of path in- 

 creases, the heat-flow must here change according to laws similar 

 to those which govern the retarding force in friction. As in 

 friction the velocities, so here the temperatures of a partition and 

 the contiguous layer of gas differ by a finite value. A diminu- 

 tion, therefore, of the heat-now must at first appear, as soon as 

 this difference of temperature commences to exert a perceptible 

 action. On further rarefaction a point is reached from which, 

 onward, the idea of a coefficient of heat-conduction loses its 

 meaning, and the flow of heat rapidly decreases as the density is 

 diminished. 



A comprehensive experimental testing of these consequences 

 has hitherto not been possible with the apparatus we have em- 

 ployed. We have, however, succeeded here in doing approxi- 

 mately what we could not in the case of friction — namely, to 

 produce a space which may be regarded as an actual vacuum in 

 relation to heat-conduction. This made it possible, also, to se- 

 parate the effects of conduction and radiation. 



In order to measure the flow of heat through a gas, we ob- 

 served (like Dulong and Petit) the cooling of thermometers of 

 various forms in glass cases of different shapes at 0°. With 

 higher pressures the action of pure heat-conduction is defaced 

 by the action of the currents formed, in consequence of gravity, 

 in the unequally heated gas. But when the pressure is lessened 

 the velocity is increased with which, under given conditions, 

 thermometric equilibrium is restored ; and therewith the influ- 

 ence of currents recedes. Thus, for a spherical thermometer, 

 cooled in a spherical glass envelope, the time of cooling (from 

 60° to 20°) between 10 millims. and 1 milliin. mercury-pres- 

 sure is independent of the latter; on the contrary, with 150 

 millims. it was once and a half as great as with 750 millims. 

 The values obtained between 10 millims. and 1 millim. mercury- 

 pressure illustrate, on the one hand, the independence of the 

 coefficient of heat-conduction of the pressure within those limits, 

 and, on the other, can be utilized for calculating that coefficient. 

 We find ;— 



Coefficient of heafc-conduction. 



Calculated by 

 Observed. Maxwell. 



Hydrogen . . 1 1 



Air .... 0-137 0141 



Carbonic acid . 0-082 0-103 



The coefficient of the conduction of heat of air in relation to 

 that of hydrogen has been already found by Stefan in harmony 



