122 PROCEEDINGS OP THE AMERICAN ACADEMY 



Investigations on Light and Heat, made and published wholly or in part with 

 appropriation from the Rumfoed Fund. 



IX. 



METHODS OF MEASURING ELECTRIC CURRENTS OF GREAT 

 STRENGTH; TOGETHER WITH A COMPARISON OF THE 

 WILDE, THE GRAMME, AND THE SIEMEN'S MACHINES. 



By John Trowbridge, Harvard University. 

 Presented, Oct. 9th, 1878. 



The measurement of electric currents of great strength can be classed 

 under four heads: No. 1. The Galvanometric Method. No. 2. The 

 Electrometer Method. No. 3. The Heat Method. No. 4. The Elec- 

 trodynamometer Method. 



No. 1. The Galvanometric Method. 



With a galvanometer of small resistance and of large radius, it is 

 necessary to bring the deflection to the neighborhood of 45° by means 

 of a shunt of very small resistance. The errors increase when the 

 deflections exceed 45° in a divided circuit, and, by the use of a shunt 

 of small resistance, any error in the measurement of this small resis- 

 tance multiplies the whole observation by this error. 



By the use of a cosine galvanometer which I devised in 1871, and 

 published in the " American Journal for Science" for that year, the use 

 of shunts can be modified ; but there are difficulties from the dip of the 

 needle and from want of accuracy in graduations of the circle which 

 measures the deflection of the moving coil from the vertical plane. 



In practice, it is very inconvenient to find a suitable shunt which 

 will answer for a wide range of experiments, and different shunts have 

 to be used. Moreover, the heating of the shunt multiplies the obser- 

 vations by an error. In short, by the use of a shunt method, we 

 measure a large quantity by observations upon a hundredth or a thou- 

 sandth part of itself, and proceed from a small quantity to a large one 

 which is a fundamentally defective method. 



No. 2. The Electrometer Method. 

 By means of a suitable electrometer, the difference of potential of 

 two points in a closed circuit can be measured ; and, from this, the elec- 

 tromotive force in volts can be estimated. The difficulty of dealing 



