312 Mr. A. K. Chapman on the 



Here the curves are no longer parallel ; the Corbino* 

 effect has a value 4187 and the Hall effect 3000 times that 

 of copper for a field of 250. The former falls off the more 

 rapidly, crosses the Hall effect curve at a field of 4250, 

 becomes zero at a field of 7900, and reaches a value of 

 — 291 at 9110. On the other hand, the Hall curve goes to 

 zero at 9325 and is -184 at a field of 10450. 



By some physicists it has been suggested that this reversal 

 is the result of the superposition of two or more effects. In 

 the usual method of evaluating the Hall coefficient such a 

 condition might arise through the simultaneous occurrence 

 of the simple Hall effect and the Ettingshausen effect ; the 

 latter giving rise to the usual galvanomagnetic temperature 

 difference at the boundaries of the plate, which in turn,, 

 because of the Thomson effect, produces a potential difference 

 which appears in the potential reading taken as a measure 

 of the Hall effect. To be sure, under such conditions,, 

 the galvanomagnetic potential difference obtained at any 

 given field depends upon the ratio between the Hall and 

 Ettingshausen coefficients, and it is conceivable that the 

 peculiar phenomena in bismuth-tin alloys might be due to 

 this circumstance. It is at once evident, though, that the 

 Corbino method entirely precludes the possibility of the 

 occurrence of the Thomson effect, since there are no free 

 boundaries present. Hence one may say with certainty 

 that the reversal obtained in the Corbino curve is not due 

 to the coexistence of the Hall and Ettingshausen effects. 



Because of the junction of dissimilar metals at the centre 

 and circumference of the disks employed, the Peltier effect 

 may enter, producing a radial heat-flow and consequently a 

 circular current. In a recent paper Zahn (An?i. d. Phys.. 

 xlvii. 1. pp. 49-82, May 11, 1915) has shown that such an 

 error in ordinary experiments on the Hall coefficient amounts 

 actually to a very small fraction of the whole, and it may 

 safely be assumed that such is the case here. 



And finally, the best guarantee that such complicating 

 effects do not play any considerable part is the reversal of 

 the radial current, which took place twenty times per second. 

 As is well known, a very definite time is necessary for the 

 Ettingshausen and Ettingshausen and Nernst effects to 

 establish themselves, since each depends upon the rise either 

 of a transverse or of a longitudinal temperature difference. 

 The brief interval between reversals is not adequate for the 

 building up of any great temperature difference. 



Nickel. — The disks were cut from a sheet of metal supplied 

 by Messrs. Eimer and Amend. 



