AND RATIO OF TIIKIR ATOMIC WKKJHTS. 7 



Tbe motion of the balance is slow, owing to the volume of water which has 

 to be ilisplaced for each vibration, but it is easy to determine within a centigi'aiume 

 the effective weiglit of the immersed system. The o])servati()ns were repeated at 

 intervals of an houi'. 



Fig. '2. — Hydrostatic weighing of.small globes. 



For tlie determination of the temperature I calibrated two thermometers whose 

 scales were divided into tenths of degrees. That one marked G has a scale of equal 

 parts; 7-1:1 had a scale intended to compensate for inequalities in the bore of the 

 tube. Both were calil)rated on the dividing engine, with no reference to the en- 

 graved scale, except in determining the proper length of the threads of mercury 

 separated. One was calibrated by Bessel's method, the other, by the method of 

 Hansen. After the calibration of the tube, the accidental errors and the inten- 

 tional inequalities of the scale were determined. The results of the calibration were 



