508 



— // and — /// is incompalible willi liiis'). I( is iiiipi'obuble that 

 tlie wire is strongly al larked over a siiuiii pari, because tlien the 

 question rises why the lesistance of the belter part o\' In — 1922 — I 

 does not disappear at the vanisiiing-])oini leinperature of liie two 

 other iiidiimi resislances. This leads to llie conceplidn tiiat llie great 

 rest-iesislance of In — 1922 — / is nniforinly distributed tiironglioiit 

 liie whole wire. The equality inter se of this quantily over the 

 three olhei' \vii-es makes this almost certain for tliem'). For the 



W 



w. 



0.04 



0,03 



0.02 



0,01 



ó lnd_1922_A. 

 □ lnd-1922_ H. 

 A lnd-19'22_III. 



3.41 



Fig 2. 



1) Measurements with hi — 1922 — A on tlie dependence of the magnetic threshold 

 value of the temperature yield for indium louglily a required field of 1,4 gauss 

 for a vanishing-point displacement of 0,02 degree, i is always 4 m. a. If in 

 agreement with Silsbee's hvpolhesis the inner magnetic field of ƒ« — 1922— ƒ 

 is to be 1,4 gauss larger than that of ƒ«— 1922— I/, the radius of In- 1922—7 

 must have been reduced to about 0,005 m.m. by oxidation, which is incompatible 

 with the ratios of the PFn'Si when it is taken into consideration that the two 

 resistances do not differ much in length. 



The variation of the resistance with different intensities of the current has not 

 been calculated in the experiments with indium wires. A current density lO-times 

 greater in a tin wire gave a vanishing-point displacement of about 0,02 degree ac- 

 cording to (Jomm. 133d, table IX. With these values for indium the oross-section 

 of J«— 19ii2 — // would have to be lOlimes that of 7w — 1922 — J, which also gives 

 a wrong ratio of the PTq's. 



') The great value and equality of this rest-rpsistaiice for all three wires made 

 us doubt the purity of the supplied indium. 



1 



