436 PROCEEDINGS OF THE AMERICAN ACADEMY 



ated circle divided into degrees, and the fractions estimated to tenths 

 by the eye. To eliminate parallax, the bottom of the compass-bos was 

 formed of looking-glass, and the eye so placed, when the reading was 

 taken, that the index and its reflection coincided. To determine the 

 constant of the galvanometer, a constant current from a thermal battery 

 was passed through it and through a beaker containing snlphate of 

 cojaper, and the weight of copper deposited measured. Two determina- 

 tions were made, and gave the result .052. To make sui-e that the 

 galvanometer followed the law of the tangents, a series of resistances 

 were interposed in the circuit, and the deflection measured. The 

 results showed that the error was extremely small, even for angles 

 as lai'ge as HC^' to 8')°. The resistance of the galvanometer was .22 

 ohms, and by it, currents from .02 to .3 vebers could be well meas- 

 ured. For stronger currents a set of shunts were prepared. The 

 wires from the galvanometer were carried parallel to each other and 

 near together for some distance to avoid their disturbina^ action on the 

 needle, and the resistance thus increased to exactly .25 ohms. Three 

 shunts. A, B, and G, were then prepared, which should reduce the 

 current to .2, .04, and .014, consisting of s-hort stout pieces of German 

 silver wire. The first and second of these were easily made by com- 

 puting their required i-esistance, and sliding them in or out of the 

 screw cups in which they were held. They were then tested by 

 I")a-sing the same current first through the galvanometer with and with- 

 out the shunt, and comparing the tangents of the deflections in the two 

 cases. To correct for the change in resistance, an additional resistance 

 Avas inserted when the galvanometer was shunted. The third shunt 

 could not be made directly, as its resistance was only .0034 ohms, and 

 we could measure directly, only to thousandths of an ohm. The 

 method of comparison alone was therefore used, reading the deflec- 

 tion when the whole current of the machine was passing, and again 

 using the o^ shunt. The correct values of the three shunts were 

 thus found to be .I'JSO, .0392, and .0107. The latter consisted of a 

 bar of German silver, .13" in diameter and 3" long. To pass from one 

 shunt to another, a simple switch or plug could not be employed, since 

 the resistance of the shunts B and C was so small that the variable re- 

 sistance thus introduced would become quite perceptible, being multi- 

 jilied many times ; and, moreover, with the stronger currents, the points 

 of contact might become heated or burnt. Accordingly, a switch was 

 inserted in the wire connected with one terminal of the galvanometer, 

 by which it could be connected with either of the three shunts, au'i a 

 second connection made with each, and with the main circuit. The 



