Gold for Weak Magnetic Fields, 83 



The reason why the values ft' differ from the absolute value 

 ft by a constant factor is to be found in the following ; — 

 v. Ettingshausen and Nernst"*, in their researches on the 

 influence of the shape of the specimen on its Hall coefficient, 

 showed that if the length of the plate (direction of primary 

 current) were not at least double the breadth (direction of 

 secondary current), the value obtained for the Hall coefficient 

 was too small ; but that it rose to a limiting value as the 

 above-mentioned proportions were approached. This they 

 showed to be due to the fact that all the current arising from 

 the transverse electromotive force does not flow through the 

 secondary leads, but that part of it leaks back through the 

 plate and the primary distributing bars. This effect only 

 becomes negligible at the secondary electrodes when the 

 -distance between them is less than half that between the 

 primary bars. 



The numbers given in column ft f of the table were obtained 

 with a square plate. The above given consideration shows 

 that these are a constant fraction of the absolute value ft. 



In fig. 3 the values of the transverse current (from the 

 fourth column of the table) are plotted against the corre- 

 sponding magnetic fields. The line represents the value 

 ft'=zo'a, the mean of all the determinations made. The four 

 highest points are omitted in order that the others may be 

 shown on a larger scale. The omitted points fall on or 

 near the same line as is shown by the values of ft' obtained 

 from them. In certain cases two or even three points fall so 

 close together on the diagram as to be undistino-uishable from 

 single points. All such cases have a small numeral, 2 or 3, 

 written near the point in question to indicate its double or 

 triple character. The diagram shows that the relation of the 

 transverse current (and hence the transverse E.M.F. in the 

 plate) is linear to the field strength. That is, that the Hall 

 coefficient is a constant over the range dealt with. The 

 apparent diminution of /3' at the lowest fields is probably due 

 entirely to errors of experiment. 



After the above given measurements were complete, the 

 plate was cut down by half in breadth, but the values of the 

 Hall coefficient were still found lower than the absolute, as 

 the part of the plate in the magnetic field was not of the 

 required proportions, and current was leaking back around 

 the edges. Slits were next cut in the foil so as to prevent 

 this, a strip one centimetre wide being left intact up the 



* Wien. Ber. xciv. (2) p. 560 (1886) ; Beibl. xi. 69, p. 352 (1887). 



G2 



