Ileat across Layers of Gas. 427 



same temperature everywhere, there is one other permanent 

 distribution possible (except at the limits), and perhaps only 

 one — that in it tliere is, for each gas and at each tension, a de- 

 finite gradient of temperature, with its accompanying equally 

 definite gradient of density in the opposite direction. These 

 results might have been arrived at in another way, viz. by a 

 consideration of the effects of the intermolecular encounters. 



Another case in which the three conditions will be fulfilled 

 is the familiar one of a imiforrii medium, in which case 



— =0, or 6'= const (y) 



But if there is a transition from one of these distributions to 

 the other, as there must be where the Crookes's layer is in con- 

 tact with the rest of the gas, there will be an interval of com- 

 promise, in which the three conditions are not strictly fulfilled. 

 (Similarly, they cannot be fulfilled where the Crookes's layer 

 adjoins the hot body A. Hence there must, in the cases that 

 really arise, be some escape of heat, which may be small, but 

 cannot vanish, because discontinuity is impossible, since the 

 length of the mean path of a molecule between its encounters 

 with other molecules is finite. Hence, also, the values of the 

 temperature at different depths within the Crookes's layer will 

 differ by small amounts from those assigned to it by equa- 

 tion (/3). It will appear, however, from the next paragraph, 

 that the rate of cooling arising from these imperfections will 

 be very slow* ; and although the heat that passes would doubt- 

 less accumulate and ultimately become considerable if there 

 were no gravity, its presence will be inappreciable in most of 

 the experiments we can make, where the portion of gas in 

 which the Crookes's layer is formed is being constantly re- 

 newed by convection currents. 



5. We have hitherto supposed that the atmosphere of gas 

 was of sufficient extent to allow the whole of the Crookes's 

 layer to come into existence ; but we shall have entirely new 

 conditions if a body B at temperature 62, which for simplicity 

 we may suppose to have a large flat surface, is placed parallel 

 to A at a distance less than the thickness of an unrestricted 

 Crookes's layer. In this case a compressed] Crookes's layer 

 will come into existence, in which, as explained in § 16 of my 

 former papers, the density of the gas must be everywhere 



* For, the Crookes's layer being in this case almost complete, the values 

 of A^, and ^6^ (see § 5) will be exceedingly small. 



t I. €. confined between a heater and a cooler, against which the layer 

 of gas expends its Crookes's stress. In withstanding this stress the heater 

 and cooler compress the layer. A compressed Crookes's layer might also 

 be called a layer of polarized gas. 



