362 COMPARISON OF RESISTANCES. 



10, ioo.... -fold by passing a gradually increasing fraction of the 

 current until all the dials are utilised; finally, when equilibrium 

 is almost obtained, the shunt is suppressed, and the deflections 

 8 and 8' on either side of zero are obtained to get the final cor- 

 rection. 



It is important only to allow the current to pass during the 

 time strictly necessary for observing the galvanometer, so as to 

 avoid heating the wires. 



In order to get rid of any want of exactitude in the ratio of 

 the arms, we may work by substitution. The resistances b and #j, 

 successively introduced between the binding screws C and B, being 

 balanced by the values n and n^ of the box, their ratio is 



965. We should be certain, especially with very delicate in- 

 struments, that the circuit of the galvanometer does not contain 

 any incidental electromotive force e (the result, for instance, of 

 a thermoelectric effect), which would considerably disturb the 

 measurements.* 



If there were only the electromotive forces E and *?, the general 

 equations (14) give 



If the condition of equilibrium is almost realised, we may re- 



place D by D (948), and the intensity by its approximate value 



a + a' 



- E; we have then 



or sensibly 



* GLAZEBROOK. Phil. Mag. [5], Vol. XL, p. 291. 1881. 



