538 Mr. L. Schwondlor on the General Theory 



and by inference 



T u 



approximately. 



Now we can decide on the method to be adopted for read- 

 justing balance. On account of the regularity-condition R= K, 

 and as both undergo variation, especially K, we are obliged to 

 adjust balance in the compensation branch by varying the 

 resistance d, and leave the coils or their armatures stationary. 



Thus the general solution of the first problem for the com- 

 pensation method is : 



1. Readjustment of balance is to be effected by a variation of re- 

 sistance in the compensation circuit, and not by a movement of the 

 coils or their armatures. By this adjustment R is kept equal to K 

 permanently, no matter in which branch the variation takes place. 



2. f=w+/3; 



a=b=g +f; 



v x=l, 



v as small as possible and X as large as p>ossible. 



ft is known from the number and nature of the single cells 

 of which the battery has to consist to produce through the given 

 line (connected up in a circuit like fig. 3) single signals with 

 sufficient strength. 



io is known from the absolute largest variation /3 may un- 

 dergo in time ; hence / is determined, and therefore also a 

 and b. 



Determination of \ and v. 



E 



We know that \v — l, and, further, that \= — should be 



selected as large as possible or v as small as possible'; but other- 

 wise it appears that no fixed values for X and v can be ascer- 

 tained. If we, however, consider the nature of the variations 

 of R and K which may disturb the balance, viz. those varia- 

 tions of R and K which are due to unavoidable decrease of the 

 internal resistance of the two batteries by the working cur- 

 rents, it will be seen that a best value of X does exist, and that 

 therefore v also becomes fixed. 

 Suppose that at a certain moment 



R=K 

 is rigidly fulfilled ; and remembering that 



R, = b + d + *, 



K = 2(a+/) + L 



