390 On Force caused between a Surface and a Gas, [Mar. 23, 



duced showed that, when making 240 revolutions per minute, the torsional 

 force on the vanes does not exceed one forty millionth part of a pound 

 acting on a lever a foot long ; that the pressure of the gas on the vanes 

 to produce this was not more than one two million five hundred thousandth 

 part of a pound on the square inch, or one thousandth part of the pres- 

 sure in a Torricellian vacuum, thus placing the extreme minuteness of 

 the forces in a clear light, a light from which the extreme delicacy of 

 Mr. Crookes's instrument had altogether withdrawn them. 



It is then shown on theoretical grounds that the difference of tempe- 

 rature on the two sides of the vanes necessary to cause heat-reactions of 

 this magnitude could not be less than l 0- 7 I\, while the probability is 

 that it is considerably more. 



In order to apply this test and see how far the actual difference of 

 temperature in Dr. Schuster's experiments corresponded with that deduced 

 from the theory, a new photometer was devised by the author with an 

 immediate view of measuring the difference of temperature caused by 

 light on a black and a white surface. 



Of two thin glass globes, 2| inches in diameter, connected by a 

 siphon-tube g inch internal diameter, one was blackened with lamp- 

 black on the inside over one hemisphere and the other was whitened 

 with chalk in a similar manner, the two clean faces of the globes being 

 turned in the same direction. Oil was put in the tube and the globes 

 were otherwise sealed up. Any light which enters through the clean 

 faces is received on the black and white surfaces, and the air in the 

 globes expands in accordance with the difference of temperature which 

 they attain, moving the oil in the tube. A motion of \ an inch on the part 

 of the oil shows a difference of 2°-2 in the temperature of the air within 

 the globes. 



The instrument so constructed is exceedingly delicate, and will show 

 a difference in the intensity of light sufficient to make one revolution 

 per minute difference in the speed of the mill. As a photometer it is 

 much more convenient than the mill, and its construction presents much 

 less difficulty. By making the lower portion of the siphon-tube hori- 

 zontal, and using glass indices after the manner of Rutherford's thermo- 

 meter, the instrument might be made to record maxima and minima 

 intensities of light, as well as be more delicate in other respects. 



Measured with this instrument, the light necessary to give the mill 

 240 revolutions per minute does not exceed 24°, and is probably less 

 than this, which shows that the theoretical difference of heat necessary 

 to cause the heat-reactions is well within the difference as actually 

 measured, leaving an ample margin for error in the methods of approxi- 

 mation used in the calculation. 



In concluding the paper the author claims to have set at rest the only 

 point respecting the explanation of the motion caused by heat which 

 remained doubtful after he had discovered that, according to the kinetic 



