142 PROCEEDINGS OF THE AMERICAN ACADEMY 



case as in the first, and the ratio of the lengths of the shunt are taken 

 to represent the ratio of their resistances. 



With a piece of uncovered copper wire, not more than two metres 

 long, and a resistance r equal to only twenty-four one hundredths 

 of an ohm, I have obtained results which are quite as good as those 

 obtained witli a box of resistance coil, using Thomson's method. 



r should be a small resistance of not over five ohms for a long-coil 

 galvanometer, and not over three ohms for a short-coil galvanometer. 

 This method offers some advantages over that of Thomson, but it is 

 not generally as good as the method due to Mance. 



