310 Mr. G. J. Stoney on Crookes's Radiometer. 



f'erence in the state of things behind the disk ; but it will cause 

 a new adjustment to be necessary in front. In fact the part 

 of the gradient which lies outside the sphere will cease to exist, 

 since there is nothing now to prevent air of the same pressure 

 and density as in the rest of the apartment from coming quite 

 up to the sphere. Accordingly, when the little instant of ad- 

 justment is over, this will be the state of things outside the 

 sphere. Hence, when the adjustment is over, there will be n 

 molecules flying inwards through every square centimetre of 

 the front hemisphere. Now the density can no longer undergo 

 change when the adjustment is over ; and accordingly there 

 must be as many molecules passing one way as the other. 

 Hence we find that after the adjustment, n molecules per 

 second must come up to every square centimetre of the inside 

 of this hemisphere and demand an exit through it, which is 

 accorded to them on the condition of their surrendering a part 

 of their momentum as they fly through the partition. For 

 since the gradient is now incomplete, the motions inside will 

 keep up an average velocity which is above v in spite of the 

 incursion of the strangers. This surrender of momentum 

 means that there is a pressure against the inside of the hemi- 

 spherical shell, which does not exist against the outside. Now 

 as the same number of molecules pass in as pass out through 

 every part of this hemisphere, we are at liberty to render the 

 shell impervious to molecules, if we substitute for its hypo- 

 thetical property the following, which can be easily attained in 

 practice, viz. the property that, although molecules may come 

 up to the shell with velocities of which the average is above v, 

 they shall rebound from it, and in rebounding shall be sent 

 off with velocities of which the average has been reduced to v. 

 This only requires a solid shell which is maintained at the 

 temperature t. This shell suffers on the outside a pressure of 

 p grammes per square centimetre, and within a pressure of 

 more than p grammes per square centimetre ; while if a simi- 

 lar impervious shell at the same temperature is substituted for 

 the hemisphere behind the disk, it suffers a pressure of p 

 grammes per square centimetre both on the inside and out- 

 side. The final change may now be made. The air outside 

 the impervious sphere may have its pressure increased and 

 become the ordinary air of a room without disturbing any 

 part of what goes on within the shell ; and we have now an 

 apparatus which is in fact no other than the radiometer with 

 its glass containing-vessel maintained at temperature t. 



17. Hence there is pressure on the inside of the front hemi- 

 sphere of glass which is in excess of the pressure on the insid< 

 of the hemisphere behind the disk. And there is of course 



