May 30, 1902.] 



SCIENCE. 



855 



■ward; after a short time it turns more 

 slowly, stops, and then rotates forward. 

 This can be continued but a short time as 

 the temperature becomes so high as to en- 

 danger tlie anemometer. 



6. Pressure .024 mm. The anemometer 

 rotated backward, but when the vessel was 

 wrapped with a non-conductor of heat it 

 rotated forward as in experiment 5. 



7. Pressure .0017 mm. The electrode- 

 less discharge was not obtained and the 

 anemometer did not rotate. Vessel re- 

 mained cool. 



8. In the experiments 1 to 7 the distance 

 from the outer edge of the anemometer 

 cups to the walls of the vessel was about 

 1 cm. Another vessel was used also in 

 which the cups were much nearer the 

 walls. "With this the backward rotation 

 was much stronger in all cases. The pres- 

 sure at which the backward rotation was 

 first obtained was much higher than in ex- 

 periments 4 to 7. 



9. In this experiment the vessel con- 

 tained a small mill which was similar in 

 construction to the anemometer, excepting 

 it had flat vanes. This did not rotate at 

 any degree of exhaustion in the strongest 

 discharge that could be obtained. 



10. A much larger vessel was used for 

 this experiment. It was 12 cm. in diam- 

 eter and the anemometer was but 3 cm. in 

 diameter, so that the distance between the 

 walls and the cups was 4^ em. The rate 

 of rotation was surprisingly great, attain- 

 ing a velocity of forty revolutions per sec- 

 ond. At no degree of exhaustion did the 

 anemometer rotate backward. This indi- 

 cates that at this great distance (4^ cm.) 

 between the walls and the cups there is no 

 radiometer effect. It is perhaps desirable 

 also to mention that a vessel was con- 

 structed having two anemometers, one 

 above the other, mounted with the convex 

 sides turned in opposite directions. At 

 the proper degree of exhaustion these ro- 



tated in opposite directions, each turning 

 in the direction of the convex side of its 

 cups. 



The backward rotation appears to be due 

 to the heat interchange between the convex 

 side of the cups and the walls of the ves- 

 sel, because: (1) By experiment 4, the 

 anemometer is acted on by a force driving 

 it backward, which persists for some time 

 after the current is interrupted. This force 

 is much less when the cups and the vessels 

 reach nearly the same temperature. (2) 

 By experiments 5 and 6, the backward ro- 

 tation is only obtained when there is heat 

 interchange between the cups and the walls. 

 The effect of wrapping the vessel with a 

 non-conductor of heat is to make the inner 

 surface of the walls nearly as hot as the 

 cups of the anemometer. When this condi- 

 tion is obtained, the backward force nearly 

 disappears and the forward force due to 

 ionic motion predominates. 



Tliis backward force appears to be a true 

 radiometer effect since it increases as the 

 vacuum becomes higher. In experiments 

 3 and 4 the distance from the cups to the 

 walls is probably greater than the mean 

 free path of the molecules and the radiom- 

 eter effect is small. In experiment 8 the 

 radiometei- effect is stronger and appears 

 at a higher pressure as the cups are nearer 

 the walls and the mean free molecular path 

 necessary for a radiometer effect is shorter. 



We may regard the molecule from which 

 the negative ion has been separated as a 

 carrier of a positive charge, and so it also 

 will be acted upon by the varying magnetic 

 field. Its velocity and amplitude will be 

 much less than that of the negative ion. 

 The amplitudes wiU be inversely propor- 

 tional to the square roots of their masses, 

 since the energy is the same in both cases. 

 If their amplitudes are of the same order 

 of magnitude as the radius of the cups, 

 the positive ions will act on the anemometer 

 also. 



