BRUSH— A KINETIC THEORY OF GRAVITATION. 181 



It is clear, then, that in the customary absence of iron anywhere 

 near the bobs they can not be unequally affected by the earth's mag- 

 netic field to any appreciable extent. 



Necessity of Very Rigid Support for the Pendulums. — When two 

 pendulums are hung from the same support, there is a well-known 

 tendency of each to influence the other by slight motions imparted to 

 the common support. It was to counteract this tendency that the 

 common support of the pendulums used in these experiments was 

 made so very rigid. 



The following experiments were made to test the efficiency of this 

 provision, and the double bobs were used because of their greater 

 weight, and the short pendulums because of their greater angular 

 amplitude. 



The pendulums were so closely synchronized that no appreciable 

 change occurred in 3 hours when swinging together. Then they 

 were started as nearly as possible in opposite phase, and again allowed 

 to swing 3 hours. Again they were started a half phase apart, and 

 allowed to swing 3 hours. In neither case was there any observable 

 change in phase relation during the three hours' run. Three hours' 

 run was deemed sufficient, because that was about the time required 

 for the bismuth to gain a whole beat when the bobs were exchanged. 



Thus it is seen that the pendulums do not observably affect each 

 other because of their common support. 



To find to what extent, if any, the pendulums affect each other 

 by the interaction of the air eddies which they set up, the bobs were 

 exchanged as usual, started together, and time required for two beats 

 gain of bismuth taken. Then they were again started, and immedi- 

 ately a large glass plate was placed vertically midway between the 

 pendulums in the line of swing, and resting on the table. Again the 

 time required for two beats gain of the bismuth was measured. This 

 procedure was repeated without, and with the interposed glass plate, 

 and with the following results : 



Time of Two Beats Bismuth Gain. 



Without glass plate 357 minutes 



With glass plate 35^ 



Without glass plate 360 



With glass plate ; 353 " 



