304 YOSHIJIRO KATO ; 



simply the result of some secondary time effect due to imper- 

 fections of our experimental method. To point out the possible 

 defects, we must first of all have a clear grasp of the meaning of the 

 term "compensation." When a current is sent through the solenoid, 

 the iron wire will be magnetised and will act on the magnetometer, 

 which will thus begin to be deflected with a certain acceleration. 

 Now, the current circulating in the compensating coil exercises just 

 the contrary action on the magnetometer. By trials, we strive to 

 make these two contrary effects on the magnetometer as nearly equal as 

 possible, which condition we can decide by looking at the image given 

 by the mirror of the magnetometer on a scale placed at a distance 

 of 1 metre. If the image does not move suddenly on pressing the 

 key, we take it to be all right. This will be quite correct, if the 

 magnetised iron and the compensating coil get their respective full 

 strengths at once. But as we know, owing to the self-induction, 

 which in the present case is not small, the current does not attain its 

 final value at once, and though the time required for attaining this 

 final state be very small, yet it will certainly be comparable with the 

 interval, during which the mirror of the magnetometer is accelerated 

 enough to produce such a small motion as our eyes can just perceive 

 by its image at a distance of 1 metre. In other words, the current, 

 and with it the intensity of magnetisation, will continue to increase 

 after the instant at which we decide the state of compensation. Now, 

 this fact, if it be present, will be of no importance so long as the 

 intensity of magnetisation increases in a simple proportion with the 

 strength of field, which in its turn increases in a simple proportion 

 with the current. In this case then the whole subsequent deflection 

 will be wholly due to the " growth in magnetisation." In fact, this 

 is just the case with a very weak field, where -^ is constant. But 

 when the field begins to get somewhat great, the curve of magnetisation 



