of Mr. Crookes's Radiometer. 397 



to the known law of steady motion, the density and pressure 

 of the fluid are greater as the motion is less. Now in the case 

 of the aetherial current maintained by the variation of atomic 

 density in the radiometer's vane, the direction will be trans- 

 verse to the parallel surfaces of the vane, and the motion will 

 be least near the black surface because, by reason of excess of 

 temperature, the channel will there be greatest. Hence there 

 will be a decrement of cetherial density from the dark towards 

 the bright surface, and the atoms, being immersed in this va- 

 riation of density, will be urged as if the vane were pushed on 

 the black surface. This agrees with the fact as observed. (It 

 is proved in art. 33 of the communication in the Philosophical 

 Magazine for September 1872, that the acceleration of an atom 

 resulting from the variation of density of the aether has a con- 

 stant ratio to the acceleration of the fluid where the atom is 

 situated.) 



Having surmised from the foregoing theory that the rotation 

 of the vanes might be affected by the streams of a magnet in 

 the neighbourhood of the radiometer, I found on trial that this 

 is the case, the rate of revolution being much reduced by the 

 proximity of a magnet of considerable power. This result 

 might be due to the influence of the magnet's streams on those 

 which, according to our theory, cross the vanes transversely, and 

 maybe taken to be evidence of the actuality of the latter streams. 

 The theory shows that the vanes are impelled by a constant 

 •accelerative force so long as the same quantity of light falls on 

 them. But from observation it appears that for a certain quan- 

 tity of incident light they attain a certain rate of revolution 

 :and then cease to be accelerated. How is this fact to be ex- 

 plained? It is evident that there must be some amount of 

 friction between the glass support and the pivot on which the 

 system of vanes turns ; and this, together with the resistance 

 which would be produced by ever so small a residuum of air in 

 the glass globe, may sufficiently account for a limit to the rate 

 of revolution. As respects friction, this explanation appears to 

 be confirmed by a peculiar circumstance which has been very 

 recently noticed — namely, that when the glass globe is caused 

 to float in water and the motion of the vanes is stopped, on 

 throwing light upon them the globe is observed to rotate. The 

 reason seems to be that the accelerative force which by the in- 

 cidence of light is made to act on the vanes, takes effect, by 

 the intervention of friction, so as to move the globe. It is 

 evident that the tendency of the friction is always to make the 

 globe partake of the rotatory motion of the vanes. I give this 

 explanation with reservation, not having yet seen a complete 

 account of this novel experiment. 

 Cambridge, April 21, 1875. 



