54 Thermal Transpiration and Radiometer Motion. 



dilation Prof. Reynolds says : — " Having thus fallen into the 

 error from which Clausius released Maxwell, the inevitable 

 result is that he arrives at the absurd conclusion that in a 

 gas at rest more molecules cross the fixed plane from the 

 cold side than from the hot side/' The absurdity appears 

 to me to be all Prof. Reynolds's, when he objects to my 

 deducing motion as the result of certain conditions in a 

 system supposed to be initially at rest, merely because it was 

 supposed to be initially at rest. I presume that what Prof. 

 Reynolds is really driving at, is the expression of his feeling 

 that he would like the motion investigated not merely at its 

 initial stage, but at a general stage, and with this everybody 

 can sympathize. But when the initial motion for given 

 permanent conditions has been calculated, it is plain how the 

 drift from hot to cold comes in in the theory of conduction in 

 gases, because as there is to be no resultant transfer of matter 

 across any area, the gas as a whole must have a drift equal 

 and opposite to that flow due to variation of density and 

 temperature, which is now not only initial, but permanent, 

 because the initial conditions are kept permanent. And yet 

 on page 148 Prof. Reynolds calmly charges it against me 

 that I fail to notice that the drift discovered by Clausius is 

 opposite to that which I consider fundamental in thermal 

 transpiration ; whereas my references to Clausius, apart from 

 their being a just acknowledgement of indebtedness, were 

 made to obviate the necessity of explaining a matter that 

 had already been cleared up as this had. On page 148 I 

 am charged with a complete misstatement of Clausius's 

 position when I say that under ordinary circumstances the 

 motion occurring in connexion with conduction in gases 

 cannot establish appreciable difference of pressure. There is 

 to my mind no misstatement here, but simply a bare state- 

 ment of fact ; consider a mass of gas at rest and at uniform 

 temperature in a cylinder whose curved surface is impervious 

 to heat, and suppose one end of the cylinder suddenly heated 

 and the other cooled so that the average temperature will 

 remain the same, then according to Clausius's theory there 

 will at first be a motion of the gas which will finally produce 

 quite a different distribution of temperature and density from 

 the original, and then a permanent state will be established 

 in which the drift neutralizes the motion caused by variation 

 of density and temperature ; but the initial and permanent 

 motions wall never be associated with establishment of an 

 appreciable difference of pressure, and I maintain that my 

 assertion of this fact is not a misstatement of Clausius's 

 position, but only a restatement in terms which bring it into, 

 its natural relation with thermal transpiration. 



