166 ELECTRICAL MEASUREMENTS 



make P, 2,500 ohms and alter the balance arms M and N to 

 correspond. 



Make M = 10 ohms and N = 1,000 ohms; gradually increase 

 P from 2,500 ohms until exact balance is obtained, with shunt 

 removed; then 



1,000 



In the above it has been supposed that with M = N the 

 deflections with P = 1 ohm and P = 5,000 ohms were one to the 

 right, the other to the left. If they had both been to the right 

 and the one with P = 1 ohm of the lesser magnitude, then the 

 proper value of P would have been below 1 ohm and X less than 

 1 ohm. In such a case proceed at once to change the ratio of 

 M and N so that 1 ohm in P balances 0.01 ohm in X. In other 



words, make -^ = T and proceed as before with the adjustment 



of P. If P should be greater than 5,000 ohms, the proper pro- 

 cedure may be decided upon from the above discussion. 



As a final precaution, all connections should be gone over to 

 see if they are tight and all the plugs firmly in place; then the 

 final balance should be taken. The bat ery current should be 

 reversed and the test repeated ; this is necessary in order to elimi- 

 nate thermo-electric currents. The average result for P is used 

 in calculating X. 



The Deflection Method. Referring again to the example 

 just discussed, with M = N, P was between 25 and 26 ohms and X 

 could be determined only to 2 or 3 per cent. Now suppose that 

 with P = 25 ohms, the galvanometer deflects from its zero posi- 

 tion thirteen divisions to the left, and with P = 26 ohms nine 

 divisions to the right; then we may interpolate, for a change of 

 1 ohm in P causes the spot of light to vary twenty-two divisions, 

 and the proper value of P for exact balance will be 25+ ohms 

 or 25.59 ohms. As the readings of the deflections cannot gener- 

 ally be taken with great accuracy, 25.6 ohms would be the value 

 of P to be accepted. It is obvious that if this procedure be fol- 

 lowed X may be determined to >^ per cent, without changing 

 the ratio from M = N. If the ratio be changed the precision 

 may be still further increased. This method is used to gain 

 precision when X is so small that P must be of small value. This 



