656 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



an approximate balance of the galvanometer forces was obtained 

 which remained fairly constant when the span of copper wire AB was 

 kept in a bath of oil to prevent sudden changes of temperature. With 

 this arrangement the following result was obtained with Plate 2 : 



Dec. 24, 1910 



59 X 10-'*^ 



Arm 0. 



13"0 



Cp. 



3.0 



H. 



5400 



The value of ^ was about 113,000, absolute, or 0.000113 ohm, and 



that of Aj? about — 3, absolute. This is a small (Quantity, and it 



may therefore be well to give 

 » a few details as to the observa- 



^^ tions. Five series, or "runs" 



were made, three with the 

 current Cp flowing south 

 through the plate, and two 

 with this current flowing 

 north. It is not safe to depend 

 on observations made without 

 such reversal of the main cur- 

 " rent ; for there is always a 

 possibility of some illegitimate 

 action of the electromagnet 

 circuit on the galvanometer 

 which can be eliminated by 

 combining two series of data, 

 one obtained with the current 



in the plate running north, the other with this current running south. 



The five series mentioned gave the following mean galvanometer 



deflections : 



Figure 13. 



With Cp north. 



+ 0.33 cm. 

 + 0.24 " 



Means + 0.29 



With Cp south. 



+ 0.11 cm. 

 -0.21 " 

 - 0.22 " 

 -0.11 " 



Final mean = ^: (0.29 + 0.11) = 0.20 cm. 



The first series with Cp south, which gives a mean deflection discord- 

 ant in sign with the others, was the shortest series and internally the 

 worst of the five. 



One other possibility of error must be considered here. The copper 

 wires Ci and Cz, being connected with the iron plate, form a thermo- 



