202 
Proceedings of the Royal Society of Edinburgh. [Sess. 
within the gap for various strengths of currents passed through the 
solenoid was measured by means of a bismuth coil. The lines of magnetic 
force due to this field ran across the nickel — that is, transversely to the 
magnetic field established in the anchor-ring coil enclosing the strip. 
The method of experiment was practically identical with the method 
used in my former papers on change of resistance with temperature. The 
nickel strip formed the greater part of one arm of a Wheatstone bridge, 
an approximate balancing being secured by adjustment of the point of 
contact on a stretched wire. The combined system of conductors forming 
the Wheatstone bridge was made part of a circuit through which a small 
steady current was passed from a secondary cell. When this current was 
flowing steadily through the circuit, one of the known resistances in the 
bridge was altered slightly in a definite manner by introducing a large 
resistance shunt in parallel with a portion of the resistance. The deflection 
obtained on the galvanometer, being due to a measurable disturbance in the 
balance, was essentially a standardising of the deflection. This calibrating 
shunt being thrown out of connection, the magnetic force or forces were now 
brought into action on the nickel strip. The balance was again disturbed 
on account of the magnetization of the nickel. 
The comparison of this deflection with the former deflection due to the 
known change of resistance in the other arm gave a ready means of calcu- 
lating the proportionate change of resistance in the arm containing the 
nickel. 
Thus let P, Q, M, N be the resistances of the arms of the Wheatstone 
bridge, N being the one of which the nickel strip forms a part. Then the 
current in the galvanometer is given by the formula 
C = E(PA - QM)/R, 
where E is the electromotive force and R is a function of the resistances 
which make up the system. In the experiments now under discussion, the 
galvanometer resistance was about 20 ohms, and the resistance of the 
branch containing the cell fully 6 ohms, while the resistances P, Q, M, N 
were all less than 1 ohm. As was shown in my former paper, this ensured 
that any slight change in M or N did not appreciably affect the value of R. 
A known change dM. in M will therefore produce a corresponding 
change in the current, whose value will be, very approximately, 
dC= - EQdM/R. 
Similarly, a small change dN in N will produce a corresponding change in 
the current of value 
dC/= -i- EP<iR/R. 
