6 Dr. J. G. Gray and Mr. A. D. Ross 



if carried out at intervals during the cooling process, would 

 result in a series of magnetization curves being obtained, 

 from which it might be supposed that the permeability of the 

 material for all temperatures lying between room temperature 

 and 800° C. might be deduced. The properties so obtained, 

 however, depend not only on the temperature, but on the 

 particular procedure adopted. The fact that the specimen 

 was tested at 800° C. determines to some extent the mag- 

 netization curve which is yielded at 700° C; in other words, 

 if the specimen had been heated to 900° C, then cooled to 

 700° C. and tested at that temperature, the magnetization 

 curve obtained would be distinctly different to that yielded 

 by the specimen at 700° C. following upon the test at 

 800° C. 



Again, if the specimen were tested at 850° C, 750° C, and 

 so on, a new set of curves would result. From the foregoing 

 discussion it will now be evident that the permeability of the 

 specimen at any one temperature for any one particular 

 field as determined from the second set of curves would 

 not in general agree with that obtained from the former 

 set. 



If, however, the specimen were submitted to a process *of 

 reversals just previous to the tests being carried out at any 

 one temperature, the results would be in strict agreement. 

 For example, if this precaution is observed a magnetization 

 curve obtained say at 400° O. following upon tests carried out 

 at 800° C, 700° C, and so on, would coincide with one 

 obtained at 400° C. without intermediate testing. 



The magnitude of the errors involved in testing a specimen 

 of annealed steel by the magnetometric method without pre- 

 viously subjecting it to the process of reversals has already 

 been pointed out. It is now proposed to investigate the 

 nature of the errois which would be introduced into the 

 results yielded by a like specimen when tested by Rowland's 

 " Method of Reversals." In this method the magnetizing coil 

 is wound with a secondary coil which is connected up in series 

 with a ballistic galvanometer. Experiment has shown that if 

 the specimen has been divested of its magnetic history in the 

 manner previously described, then, on reversing in turn in 

 the magnetizing coil each of a series of currents of increasing 

 magnitude, the inductions calculated from the corresponding- 

 kicks of the galvanometer give, when plotted against the 

 magnetizing forces in the coil, a curve which coincides with 

 the true magnetization curve for the specimen ; that is, the 

 curve which would be obtained if the specimen were subjected 

 to the process of reversals previous to the establishment of 



