536 PKOF. C. G. KNOTT ON 



the other ; and this whole compact arrangement could be lifted in and out of the air- or 

 oil- bath in which the nickel wires were raised to different temperatures. The two 

 anchor rings lay close together, and the four terminals of the pairs of coils were led up 

 the side of the pillar and connected to commutators so that either, neither, or both 

 nickels could be magnetised or not by the passing current as occasion required. From 

 the upper ends of the nickel rods nickel wires were led to form the other arms of the 

 Wheatstone Bridge and give the necessary connections to the galvanometer and to the 

 single cell used for testing the resistance. It was advisable to make the connections 

 with nickel wire so as to prevent as far as possible thermoelectric currents. 



The resistances which formed the other arms of the Wheatstone Bridge will be 

 distinguished as m andn; and the balance of the galvanometer was produced when 

 Mw = Nm. 



The battery branch joined the junction of M and N to the junction of m and n ; 

 and the galvanometer branch passed from the junction Mm to the junction Nn.- A small 

 part of m was measured off and connected in multiple arc, when necessary, with a variable 

 resistance R supplied by a resistance box of the dial construction and capable of giving 

 resistances from 1 to 10,000 ohms. By this means the resistance of to could be rapidly 

 adjusted within certain limits to any required value. I shall call the part of to so used 

 as a shunt with R the resistance x. 



At the beginning of each day's experiment, and occasionally throughout the experi- 

 ment, the galvanometer readings were calibrated by measuring the deflection produced 

 by a slight change in the resistance to. Suppose, for example, that the deflection pro- 

 duced on reversal of the galvanometer circuit was a when the shunt resistance was R, 

 and that it became a' when the shunt was R', then the change of resistance correspond- 

 ing to the change of deflection a — a' is easily shown to be measured by the quantity 

 x 2 ((x + R)~ 1 — (x + R')~ 1 ). Since the cell used was a secondary battery of very small 

 resistance acting through a resistance of 10 ohms, we could safely assume the electro- 

 motive force to remain constant for one day's experiment. Indeed, day after day with 

 the adjusting magnet of the galvanometer in the same position the calibration experi- 

 ments gave the same results to within a half per cent. 



In the final series of experiments, the coils were immersed in olive oil, which was 

 kept at a constant temperature by being surrounded by a second vessel filled to a sufficient 

 height with water, whose temperature was kept steady by a suitably adjusted Bunsen 

 flame. In this way very satisfactory observations were obtained at temperatures 12°, 

 57°, and 93° C. The attempt to reach higher temperatures in this series of experiments 

 foiled, for it was found that the oil began to conduct at about 120° C. Moreover, since 

 in the earlier experiments in the air-bath the temperature had once or twice been taken 

 above 160°, the silk covering on the outside wires began to give way; and finally the 

 insulation broke down first in the coil M and then in the coil N. That is to say, the 

 nickel core came into electric contact at some one or two points, and the comparatively 

 strong magnetising current gave rise to a difference of potential in the one arm of the 



