UPON MAGNETISATION, AND CONVERSELY, IN IRON, STEEL, AND NICKEL. 517 



point, now well established, after a cyclic extreme is departed from, and which may 

 even occur close to the vertical axis, disposes of any connection between the value of 

 dB/clH. at any point and the magnitude of the spark effect. The spark effect, or the 

 difference B v — B when vibrations are used, may be a simple function of dB/dH. at 

 any point when the B, H curve is determined by increments' from zero ; but this 

 cannot be so during the cyclic process. 



Dr Eccles' experimental method reproduces to a certain extent the working 

 conditions of the first form of Marconi's magnetic detector of electric waves, and 

 confirms the latter's experience that the magnitude of the spark effect is small as a 

 cyclic extreme is departed from. But in neither case are oscillations superposed upon 

 a hysteresis loop which has not already been modified by the effect of previous 

 oscillations. In neither case are the A conditions fulfilled. 



