128 PROCEEDINGS OF THE AMERICAN ACADEMY 



sion of cylinders from axis ; r = radius of cylinders ; and w, mass of 

 moving parts before w was added. The dimensions being in milli- 

 metres. 



From these expressions, we obtain 



TT 2 k 



The constants G and g were calculated from the actual measure- 

 ments of the coils, which could be made with great accuracy, since all 

 the parts were large. 



The constants were as follows : — 



mean radius r = 153.3 mm. 

 Gg = 1631.45 



^ = 656.626. 



The constant was also determined by the running the same current 

 through the electrodynamometer and a tangent galvanometer of one 

 turn of copper wire, whose radius was r and whose constant was 



equal to - — • 



2/17T 



T . . ~ 2 r*T2 . ia F . ,F r*T* tan20 



In this case C z = , ., „ tan* 1 a = ^- sm qp, and tt- = . „ , • - — , 



4«-ir- h,fj x (jg in'Tr* sin <£' 



where T= horizontal form of earth's magnetism. 

 r = radius of galvanometer coil. 

 6 = deflection of galvanometer, 

 qp = deflection of electrodynamometer. 



The result obtained in this way closely agreed with that obtained by 

 the previous method. 



No difficulty was experienced from the heating produced by currents 

 of even eighty vebers, when the current was allowed to run for a long 

 period through the instrument; as long as the stream of water was 

 maintained around the mercury cups, even a small immersion of the 

 points of the axis of the movable part of the instrument did not result 

 in heating. By this instrument, therefore, the whole current could be 

 measured without shunting. At first, the metal pivots which dipped 

 into the mercury were tipped with aluminum ; but, when a strong cur- 

 rent passed through them, the mercury was disturbed by an apparent 

 ebullition, and the mercury speedily was covered with a black deposit. 

 It was found that copper points would answer perfectly well. Dis- 



