182 BRUSH— A KINETIC THEORY OF GRAVITATION. 



The mean effect of the glass is to speed up the bismuth 6 minutes, 

 or about 1.7 per cent. 



This small discrepancy in periods may be largely due, and almost 

 certainly is partly due, to the effect of the glass plate on the slight air 

 currents unavoidably present in the room. The plate shields one of 

 the bobs from these currents, and by reflection accentuates their effect 

 on the other bob, except when, if ever, the currents happen to be 

 closely in the plane of the plate. 



Hence we may conclude that the effect of interacting air eddies set 

 up by the bobs is negligibly small. 



The foregoing review and extension of the pendulum experi- 

 ments, unless some at present unsuspected source of large error is 

 discovered, leads me to believe firmly that there is a real and very 

 considerable mass-weight difference between zinc and bismuth. This 

 belief is strongly supported by the torsion pendulum experiments of 

 last year, and by several other lines of experiment now in progress, 

 one of which I shall briefly describe : 



A most carefully designed and constructed instrument of precision 

 has been built for comparing the velocities of freely falling bodies in 

 two aluminum containers, alike in size, shape, and smoothness of 

 surface. These are loaded to exactly the same weight, to equalize air 

 resistance effects, and dropped simultaneously, side by side, from 

 exactly the same height always. Have had no trouble with this part 

 of the apparatus. 



After falling about 120 cm. the containers strike a pair of targets 

 exactly equidistant from their starting points. This adjustment is 

 made by loading the containers with the same metal and reversing 

 their position again and again while adjusting the targets until the 

 containers strike them simultaneously. The mechanism is such that 

 the containers push the targets aside after striking them. 



When the containers are loaded with zinc and bismuth, the one 

 containing bismuth strikes its target just a little before the zinc one 

 arrives. The time difference must be of the order of a hundred 

 thousandth part of a second, and the beautiful timing mechanism 

 detects this easily. 



But in using this apparatus a persistent collateral phenomenon 

 developed, which promises to be of great interest. This is being in- 



