

ANOMALOUS BEHAVIOUR OF DELICATE BALANCES, ETC. 415 



Summary. 



The experiments detailed in this paper indicate that the majority of delicate 

 balances are liable to be adversely affected (1), by changes in the temperature of the 

 neighbouring air and (2), by certain slight inherent, and possibly unavoidable, 

 imperfections in the structure of the beam. 



(1) The errors due to the first-named cause may be introduced in three different 

 ways : 



(a) By a relatively large but very unequal increase in the temperature of the two 

 arms of the beam during the operation of loading (Fig. XII.). The error arising from 

 this may be avoided by enclosing the beam within an extra case of metal and glass 

 (Fig. X.). In the absence of a protecting case, an interval of from 10 to 15 minutes 

 should be allowed to elapse between loading and the final determination of the 

 weight. 



(6) From a want of symmetry in the position of the balance with regard to the 

 radiant energy entering by the window and emanating from the walls of the room 

 (Curve II, Fig. XIII. ). 



The best position for a balance in a given room can only be determined by trial. 

 When a suitable differential bolometer is available very little difficulty need be 

 experienced in selecting the most favourable spot. 



(c) By any strictly uniform change in the temperature of the whole beam. 



If the balance is not kept at one temperature it will be necessary, in order to 

 secure the highest possible degree of accuracy, to know the II. P. temperature 

 coefficient for any given load. It may be convenient to determine once for all the 

 R.P. temperature coefficients for some three or four suitably chosen loads; the 

 corresponding values for other loads can then be found with sufficient accuracy by 

 interpolation. 



(2) The errors which may be introduced through slight imperfections in the beam 

 may be evaded by first counterpoising the object to be weighed with the requited 

 standards and then leaving the beam to oscillate for about 10 minutes. By this 

 device the beam is fatigued and reduced to a normal condition. The actual weighing 

 may then be accomplished. 





