^-^ Mr. G. J. Stoney on the Penetration of 



greater than at the same distances from A in the complete 

 Crookes's layer, to preserve the lateral pressure unchanged. 

 Through each element 8S of an isothermal surface the mole- 

 cules will still travel in equal numbers inwards and outwards, 

 because when the adjustment is once over, the density of the gas 

 will not anywhere undergo further change ; but the molecules 

 making their way outwards (/. e. from A towards B) will, on 

 the whole, be swifter than those tending inwards, because 

 there should be a complete Crookes's layer to enable the swifter 

 class of molecules rebounding from A to keep back the whole 

 of the slower kind which constantly tend to crowd in (see 

 Phil. Mag. April 1876, p. 308, §§ 15, 16, and 17). Accord- 

 ingly, if the molecules at any one moment within an element of 

 volume be considered, the portion of them which form a pro- 

 cession travelling inwards will now be found more numerous 

 than those advancing outwards, and at the same time so much 

 slower that the momentum in the two directions is the same ; 

 in other words, there is no molar motion of the gas, nothing 

 in the nature of a wind. But that there is a continual transfer 

 of kinetic energy from A to B across the intervening gas is 

 evident, because members of the procession of colder molecules 

 crowding up to A will cause the temperature (9i — A^i of the 

 inner surface of the Crookes's layer to be lower than 6i, the 

 temperature of A ; while, at the same time, the members of the 

 swift procession which reach B will cause 62 + A^g? the tem- 

 perature of the outside surface of the Crookes's layer, to be 

 warmer than 62, the temperature of B. The Crookes's layer, 

 accordingly, must acquire heat by its contact with A, and im- 

 part heat where in contact with B ; and as adjustments within 

 the layer are made with a speed comparable with the velocity 

 of sound in the gas, it is possible to arrange experiments in 

 which the differences of temperature A^i and A ^2 shall have 

 any amounts from — when the interval between A and B 

 equals or exceeds the width of an unrestricted Crookes's layer, 

 — up to values bordering upon i(^i — ^2) — which, in the cases 

 where the temperatures 6^ and 62 are not far asunder, is close 

 to the limiting value produced by diminishing the interval be- 

 tween A and B, or by attenuating the gas. 



Accordingly, if the variations of temperature were plotted 

 down on a diagram, the ordinates representing temperatures, 

 and the abscissas distances measured perpendicularly to the 

 isothermal surfaces within the gas, we should obtain a figure 

 something like that on next page. It is moreover manifest 

 that the curve m n, representing the variations of temperature 

 across the compressed or polarized Crookes's layer, will ap- 

 proximate more and more to a horizontal line the greater the 

 tenuity of the gas. 



