Transpiration and Radiometer Motion. 53 



(or the walls themselves) from any part in thermal tran- 

 spiration, which he rests solely on the action of direct 

 conduction of heat." Now I do no such thing : what I 

 really say is, that the solid walls dominate the temperature 

 of the gas so completely that the temperature of the gas in 

 any section of the tube may be taken as that of the solid 

 walls bounding it ; then suppose this distribution of tem- 

 perature maintained in any way, and the solid walls may be 

 supposed to be abolished as far as the calculation of the 

 number of molecules crossing any section is concerned. Yet 

 although the action of the walls on the gas through viscosity 

 is subsequently treated as of fundamental importance, 

 Prof. Reynolds lays this simple little mathematical artifice of 

 supposing the walls abolished for the purpose of minimizing 

 the verbiage of a certain calculation, to my charge as a total 

 ignoring of the walls. It is quite plain that I do not rest 

 thermal transpiration " solety on the action of direct con- 

 duction of heat,'"' if Prof. Reynolds means by this the normal 

 conduction in an indefinite mass of gas ; I take account of 

 the action of the walls both when I ascribe their temperature 

 to the gas, and when I consider their frictional resistance to 

 the motion of the gas. 



Second : — On pages 144 and 145 Prof. Reynolds gives an 

 account of Clausius's discovery that the colliding molecules in a 

 gas conducting heat must be considered to have a uniform drift 

 from hot to cold, because, while equal numbers of molecules 

 are travelling in opposite directions across any section, those 

 crossing from the hot side are moving more swiftly, and he 

 then seeks to imply that I have made the " somewhat common 

 and certainly historical error in mechanical analysis " of 

 neglecting this drift. On page 147 he states of me : — "he 

 gives reasons for thinking that the walls of the tube produce 

 no effect on the characteristics of the molecules which rebound 

 from them. And thus having cut away the only cause of 

 thermal transpiration he proceeds : " I have already pointed 

 out what a complete misstatement this is of my contention. 

 In my paper I gave a simplified and brief calculation, after 

 Clausius, of the number of molecules that must cross unit 

 area in unit time in a gas at rest on account of a given density 

 and rate of variation of density and a given temperature and 

 given rate of variation of temperature across that section. I 

 did not assert that the gas was to be at rest to begin with, 

 but as it was not stated to be moving, that was assumed, and 

 I only calculated in the usual way, only more briefly, what 

 must be the number crossing unit area in unit time so long 

 as the given conditions are maintained. Of this little cal- 



