322 Mr. A. K. Chapman on the 



inductance is plotted against the field, the same peculiar 

 maximum occurs, although the field at which it takes place 

 is different and the absolute value of the peak is altered. 

 In all probability the occurrence of the maximum at a 

 distinctly lower field is due to the fact that, when the magnet 

 was used, one of the pole-pieces was traversed by a hole, 

 which so distorted the field that the density of lines was very 

 low over the central part of the plate. It is evident from 

 the formula for the circular current, 



C= Q -log^IH, 



that its magnitude is inversely proportional to logl/^j, and 

 hence, if the region of small radii is excluded from the field, 

 that a proportionately stronger field is needed to produce a 

 given circular current. Obviously, the field over the whole 

 disk is uniform in the case of the solenoid, and so, since the 

 region of small radii comes under its full influence, it might 

 be expected that a given peculiarity in the curve would take 

 place at a lower field strength. 



On pursuing the investigation to values of H below 300, 

 another remarkable effect was brought to light, as is shown 

 by the same curve. Below this point the circular current 

 is negative. So far as the writer is aware, this is the first 

 instance in which such a change in sign in the Corbino 

 effect has been detected in a pure metal. At first sight it 

 seems reasonable to suppose that this reversal might be due 

 to residual magnetism, but such is not the case. A current 

 of 33 amperes was passed through two coils of the solenoid 

 connected in parallel in such a way as to produce a positive 

 flux, then readings of the Corbino effect at a positive field 

 of 100 were obtained and found to be negative. Next, an 

 equally high negative field was applied for a time and a 

 balance again sought at 100 positive. The Corbino effect 

 was negative as before, showing that the reversal is not 

 due to residual magnetism, for the current of 33 amperes was 

 considerably higher than any actually used for the plots 

 given. It should be remarked though, that, after taking a 

 set of determinations for a curve, going from low to high 

 fields, the disk was always carefully demagnetized by reversals 

 before proceeding again to lower fields. This was purely 

 a matter of precaution, as the effect of residual magnetism 

 was actually very small. 



In fig. 9 we have a plot of the circular current divided by 

 the field ; here the maxima and minima are very sharply 



