ILLUSTRATING CARRIAGE OF MOMENTUM 55 



Thus in a particular experiment the observed 

 torque was 21 x IO" 6 cm. dynes, while the torque 

 calculated from the energy was 22 x ICT" 6 cm. 

 dynes. But there was doubtless considerable gas 

 action and the close agreement is probably acci- 

 dental. All that can be said is that the calculated 

 and observed effects agreed within a few per cent. 1 



When the beam was directed on to the silvered 

 disc, the deflection, as might be expected, was 

 much less. For the reflected beam took away the 

 momentum parallel to the surface which the 

 incident beam brought up. 



To obtain consistent results this form of appara- 

 tus requires very accurate construction and adjust- 

 ment. For if the black disc is not accurately 

 vertical, if the normal through its centre does not 

 pass exactly through the axis of suspension, and 

 if the incident beam is not perfectly uniform all 

 over the disc, the disturbing forces due to con- 

 vection and radiometer action may easily turn the 

 suspended system more than the light-force and 

 quite possibly in the opposite direction. 



Another form of the experiment was adopted 

 which more easily gave definite and consistent 

 results. A blackened disc of mica, about 5 cm. in 



i The values here given and those in the succeeding ex- 

 periments are the results of a redetermination of the various 

 constants and a revision of the calculations, and are not quite 

 in accordance with the values given in the papers cited in 

 the footnote above. 



