374 Mr. E. T. Glazebrook on a Method of Comparing 



which leads to the condition that, when there is no throw of 

 the galvanometer, R / C / = EC. 



We proceed to inquire what resistances will give the most 

 accurate value for the capacity C in terms of CK, the known 

 capacity of a standard condenser when using a given galvano- 

 meter. Let us suppose the adjustment made by varying E, 

 and determine the error 8a produced in a by an error SE in E. 

 Then, remembering that, when the adjustment is perfect, 

 EC = E / C / and a=Q, if 8~R is the error from perfect adjust- 

 ment, we have 



6a -""HT(G + E + E') ; • • • • W 



E / C / 

 and if 8G is the error in the capacity, since = ^ , 



an - _ R/ ^R _ HT(G + E + E / )E / C / i , 

 6{j ~ ~~BF~~ 2Ett£CE 2 da; 



or, since CR=C'R', 



HT(q + E + EQ 

 SG ~ 2E^E Ba (21 > 



Now k varies as the number of turns in the galvanometer, 

 and so also does G ; 



.-. K=^G, 



• • SC -2E^lG + E + GEi J * * (22) 



and if we suppose that we are liable to an error 8a in a, the 

 error in C is least when the resistances E and E' are both 

 high. 



Thus it is best to use, with a given galvanometer, high 

 resistances E and E r . 



We arrive at the same result if we make the adjustments by 

 varying E / instead of E. 



Again, let us suppose that we have a galvanometer with a 

 given channel, and we wish to fill it with wire so as to be most 

 sensitive. Let V be the volume of the channel, y the radius 

 of the wire, I its length, p its specific resistance, and suppose 

 we neglect the thickness of the covering ; then 



a - iL - M 2 



Try 2 " V ' 



k=c/ P , 

 where g depends only on the form and dimensions of the 



