The Mechardcal Action of Eadialion. 



7 



overwhelming power of convectioii currents is overcome 

 by perfecting the vacuum. 



But there is anotlier j>hase of this curious action which 

 seems to have escaped the notice of former observers. It 

 may be described as repulsion in air. The appHcation of a 

 very gentle heat will, under certain conditions, drive the 

 disk away from the warmed surface, even in air of ordinary 

 density. 



At intervals, from the time of my earliest experiments, 

 this puzzling motion would thrust itselr before me until I 

 was convinced that it was not an experimental accident 

 but that it was a legitimate effect of some rare combination 

 of conditions. What these conditions are I set myself to 

 discover. 



It was an effect which I could not, at first, produce at will. 

 Sometimes it would appear in the early morning but refuse 

 to be reproduced as the day advanced. Sometimes it 

 would occur in the eveninor wlien no trace of it had been 

 seen during the day. A day of alternate showers and sun- 

 shine seemed to be most favorable to its production. 

 Kemembering that the walls of the chamber were to a 

 considerable degree impervious to heat so that the tempera- 

 ture within suffered no rapid changes as did that without, 

 these facts suggested the direction in which to seek for the 

 cause of the action. By inserting a thermometer into the 

 chamber and placing another outside, the difference of 

 temperature between the interior of the instrument and the 

 air outside could be measured and its rehition to the motion 

 of the needle could be studied. A multitude of observa- 

 tions followed. In every case when the repulsion occurred 

 the temperature of the interior of the chamber was found 

 to be higher than that of the external air. With a difference 

 of a single degree (F.) the needle would be repelled by tbe 

 gentle heat of the hand held at a distance of twelve inches, 

 while by a somewhat stronger heat the motion would be 



