BRUSH— KINETIC THEORY OE GRAVITATION. 49 



The attraction of B on the silver balls must draw the oscillating 

 system out of plumb; but as this effect amounts to only about a 

 millionth of a millimeter, it is entirely negligible. 



Prior to using the apparatus the cylinders are swung into a posi- 

 tion at right angles to that shown, all lights are extinguished except 

 those to be used in making observations, and the room is closed 

 against all draft. After two or three hours of repose, to equalize 

 temperatures, one of the cylinders is moved into operative position 

 a minute or two only, to start a definite movement of the oscillating 

 balls. Then the diminishing oscillation limits are read 6 or 8 times 

 to establish the zero point. Next, one of the cylinders is moved into 

 operative position and left there until 6 or 8 oscillation limits have 

 been read ; then the cylinders are exchanged by reversing the carrier, 

 and 6 or 8 more deflection readings are made. Twice again the 

 cylinders are exchanged and similar readings taken, so that two sets 

 of readings are had for each cylinder. Usually they agree very well 

 indeed, and their mean is taken as the true value. Finally, both 

 cylinders are swung out of position as at first, and another set of 

 readings taken to redetermine the zero point. 



Although such a series of readings occupies about four hours, 

 the zero drift rarely exceeds two scale divisions (2 mm.) and, as- 

 sumed to have been progressive, is apportioned among the several 

 sets of readings. 



In the above manner many comparisons have been made of lead 

 and bismuth ; lead and zinc ; bismuth and tin ; bismuth and zinc ; 

 silver and zinc; and lastly of aluminum and bismuth and aluminum 

 and zinc. 



The interlacing observations support each other very satisfac- 

 torily. 



In every case the observed deflection is divided by the weight of 

 metal causing it, so as to reduce all to a common standard of attrac- 

 tion per kilogram. 



Table I embodies the results thus far obtained, taking zinc for a 

 standard and calling its attraction per kg. 100. 



