Mr. G. J. Stoney on Crookes's Radiometer. 313 



value of this fraction will scarcely differ from unity when p 



becomes loo^oo or 1,000,00 °f an atmosphere. Using these values 



for — £— , we can compute by the foregoing formula how much 



the front of the disk would need to be warmer than the rest 

 of the apparatus at these several tensions, to produce a Crookes's 

 pressure of the hundredth of a milligramme per square centi- 

 metre, which is an amount Mr. Crookes observed. 



24. We thus find that if there were no convection-current we 

 should produce this pressure by warming the front of the disk 

 about O09 of a degree at a tension of j^- of an atmosphere, 

 while an elevation of 0*6 of a degree would be required at a 

 tension of IO q 000 , and an elevation of 6° at a tension of - 000 000 of 

 an atmosphere. This last excess of temperature may be looked 

 upon as inadmissible, since it is not likely that the front of the 

 disk is ever so much as 6° warmer than the back. Hence we 

 may, I think, conclude that where Crookes's pressure is con- 

 spicuously* present, the exhaustion does not approach the 

 millionth of an atmosphere t. 



25. The probability seems to be that the effects of convection- 

 currents are sufficiently feeble at tensions between j^- and 



100,000 of an atmosphere for us to fix on some tension between 

 these limits as the tension at which we are likely to find the 

 greatest apparent effects, i. e. the greatest excess of Crookes's 

 pressure over the resultant of the forces due to convection- 

 currents. 



* It must, however, be remembered that a force very much less than 

 the hundredth of a milligramme per square centimetre would manifest 

 itself by causing motion in a radiometer. 



t This would afford a means of ascertaining whether there is any re- 

 sidual effect not accounted for by the foregoing explanation, if we might 

 depend on the methods by which very low tensions are determined. But 

 it is evident that a gas which can be occluded in any of the solids present, 

 or absorbed in any liquid, or condensed upon the surface of either, or any 

 trace of vapour which either the solids or liquids can give off, will at a 

 certain point of tenuity elude detection. All direct methods are on this 

 account so untrustworthy that I should be disposed to think that low 

 tensions may be best estimated by observing the amount of torsion 

 which the Crookes's pressure in them can give to a fibre of glass or silk, 



and compaiing it with the formula in § 22, using unity for — x— when the 



tension is excessively low. A gauge might easily be constructed upon 

 this principle. 



