Circular Magnetization on Magnetized Iron Wire. 425 



by successive tests before and after heating. Before pro- 

 ceeding further the zero-point of the specimen was again 

 tested, and was found to agree to within 1/10 of one percent, 

 with the previous determination. There was no trace of 

 oxidation. 



The Observations. 



The method of taking the observations was as follows : — 

 First, the containing tube was exhausted. The magnetometer 

 deflexions were observed at longitudinal fields ranging from 

 1 to 30, with currents in the wire varying from to 1 6 amp. 

 It was previously observed that the wire was practically 

 demagnetized at 16 amps., which corresponds to a temperature 

 of 750° C. 



The current in the solenoid was reversed several times 

 before each reading of the magnetometer deflexion, thus 

 ensuring a reversal curve. The current in the wire was kept 

 constant for each reversal curve. Its value with that of the 

 corresponding resistance of the wire was observed at intervals 

 along the curve. The effect of residual thermal currents in 

 the heating circuit was eliminated by reversal of the current 

 in the wire. 



The Sprengel pump was kept running during the whole 

 set of observations in order that the gases given off from 

 the heated iron and copper, as well as air which might leak 

 in owing to imperfect sealing, might not affect the vacuum. 

 Under these conditions the vacuum was kept very high and 

 constant, and the iron wire remained bright throughout the 

 whole series of tests. 



On the completion of this set of observations the vacuum 

 was let down, the containing tube disconnected from the 

 pump, and a similar set taken in air. All the conditions 

 remained the same as before, except that the wire was tested 

 in air instead of in a high vacuum, and was therefore at a 

 necessarily lower temperature for the same heating current. 



As soon as possible afterwards, two similar series of obser- 

 vations were taken at much higher fields, varying from 50 to 

 300. The conditions were exactly the same in this case as in 

 the lower fields, except that the controlling field of the 

 magnetometer had to be strengthened, and that the vacuum 

 at which the higher temperature observations were taken was 

 slightly less perfect. 



At this stage the zero-point of the specimen was again 

 tested and was found to agree with the two previous deter- 

 minations, within the limits of accuracy of the method. It is 

 interesting to observe that the electrical resistance was 



