320 



and thus renders easy a degree of accuracy in the measurement of 

 resistances quite unattainable with any other form of Wheatstone's 

 balance. In the new method which I now propose for low resist- 

 ances, I make the secondary conductor on exactly the same plan, 

 and generally of about the same dimensions, as the primary. The 

 bisected testing-conductors are only available when the resistances 

 of the standard and of the tested conductor can be made equal ; and 

 with them the method which has been described above seems to be 

 the most accurate possible for testing a perfect equality of resistance 

 between two conductors. 



The same plan of testing-conductors seems still the best, even 

 when testing by equality cannot be practised, with only this differ- 

 ence, that the two branches of each testing-conductor, instead of 

 being made of equal resistance, must be adjusted to bear to one 

 another very exactly the ratio which the tested resistance is to bear 

 to the standard. By proper care, to prevent the bobbin of either 

 testing-conductor from getting any non-uniform distribution of tem- 

 perature, great accuracy may still be secured; but it is scarcely 

 possible to maintain so very close an agreement of temperature, 

 and therefore so constant a ratio of resistances, as when the two 

 branches are equal lengths of one wire coiled side by side. 



The use of this plan of conductors divided in a fixed ratio, whether 

 for the single testing-conductor in "Wheatstone's balance, or for the 

 primary and secondary testing-conductors in the new method now 

 proposed, requires that either the standard or the tested conductor 

 can be varied so as to adjust the resistance of one to bear precisely 

 that ratio to the resistance of the other. In certain cases this may 

 be done advantageously by shifting one or other of the contacts 

 S, S', T, T' along the standard or the tested conductor, as the case may 

 be. If, for instance, T or T' can be shifted conveniently, the object 

 of the measurement may be to find by trial on the tested conductor 

 a portion TT' from mark to mark, of which the resistance bears a 

 stated ratio to the fixed standard SS' from mark to mark. But by 

 far the easiest working, and in most cases the most accurate also, is 

 to be done by means of a well-arranged series of standards with 

 terminals adapted for combining them in such a manner as to give to 

 a minute degree of accuracy whatever resistance may be required. 

 In a future communication on standards of electric resistance, I 



