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



momentum somewhat further. In this way the characteristic 

 of each procession is retained over a considerable distance, 

 although it is gradually effaced as the procession advances. 

 As the swift procession advances it becomes less swift ; as the 

 slow procession advances it becomes less slow. If there is 

 room between the disk and the glass for the entire gradient, 

 that is, for the whole thickness of those layers which I have 

 described, then that part of the slow procession which reaches 

 the disk will have not only lost its sluggishness, but will have 

 attained the full molecular speed of the beginning of the swift 

 procession ; and the swift procession, where it comes into con- 

 tact with the envelope of glass, will have become as slow as 

 the beginning of the slow procession. In such a case there 

 would be no Crookes's pressure. But if there is not room for 

 the full gradient, then each procession will retain a portion of 

 what characterizes it when it reaches its destination. The front 

 of the swift procession, where it reaches the glass, will still re- 

 tain an average of molecular motions more rapid than that 

 which corresponds to the temperature of the glass ; and that 

 part of the front of the slow procession which reaches the disk 

 will have its internal motions less active than those correspond- 

 ing to the temperature of the disk, while molecules of a still 

 more sluggish type will alone reach the back of the disk and 

 the hemisphere of glass behind the disk. 



15. Here, then, we have conditions which would produce a 

 difference of pressure on the front and back of the disk, if 

 equal numbers of molecules had access to both sides. How- 

 ever, the number striking the disk in front will not be quite so 

 great as the number striking its back, because the members of 

 the swift procession will shove back some of those which other- 

 wise would have made their way up to the disk. But they do 

 not keep back enough to prevent there being some difference of 

 pressure ; for it would require the whole gradient to produce 

 such a defect of density as would entirely compensate for the 

 greater average momentum imparted by each molecule that strikes 

 the front, as compared with the momentum imparted by each 

 molecule striking the back of the disk. This is the cardinal 

 point in the explanation. 



16. As this is the key to the whole explanation, and as I 

 find that some who feel an interest iii the subject find a diffi- 

 culty in apprehending it, I will give a proof of it from another 

 point of view, and with every necessary detail. Let us imagine 

 a large apartment containing throughout air as attenuated as 

 the air in the radiometer. Let its temperature and the tem- 

 perature of the walls of the room be t, and its tension p 

 grammes per square centimetre, p being a sufficiently low frac- 



