32 



Mr. G. Barlow. Effects of Magnetisation [June 18, 



and Jones, and theoretically it should eliminate the temperature effects 

 due to (1) the Joule heating effect in the iron wire caused by the 

 current which flows through it, (2) radiation and conduction of heat 

 from the magnetising solenoid, (3) gradual changes of temperature in the 

 whole apparatus. It was found, however, that the second of these 

 effects still gave rise to serious difficulties, since the necessary insula- 



Fig. 1. 



G 



tion between the coils P and Q prevented rapid equalisation of tem- 

 perature. Thus the transmission of heat from 'the solenoid affected 

 the spiral coil Q sooner than it affected the coil P, and hence the equi- 

 librium of the bridge was disturbed. This method was therefore 

 modified, the following arrangement being finally adopted. It will be 

 sufficient to describe only the experiments on nickel, as in the case of 

 the other two metals the apparatus was essentially the same. 



The resistance P consisted of two complete turns (i.e., four lengths) 

 of the nickel wire wound longitudinally on a thin rod of wood 65 cm. 

 long. The connections to the bridge were made of thick copper wire 

 soldered to the ends of the nickel coil. The resistance Q was equal to 

 P, but consisted of thin copper wire wound in a similar way on the • 

 same rod. All the wires had double silk insulation. The coils were 

 then covered with two layers of white wool and placed in a long glass 

 tube which fitted into the magnetising solenoid; also the latter was 

 provided with an internal water jacket, through which flowed a stream 

 of cold water from the mains, the temperature of the water on enter- 

 ing and leaving the solenoid being indicated by two thermometers 

 inserted in capsules through which the water flowed. This arrange- 

 ment prevented rapid transmission of heat to the coils P and Q, so 

 that the heating effect was always small, and varied very slowly ■ 

 moreover, the heating effect in Q sufficiently compensated that in P. 

 The current in the nickel wire never exceeded 0*05 ampere, and the 

 bridge circuit being only closed for very short intervals, this current 

 caused no observable change of resistance through the Joule heating 

 effect. 



