( 74) 



long, when two opposite borders of the plate were maintained res- 

 pectively at 100° and at the temperature of the room, in a magnetic 

 field of 9500 C. G. S. a difference of temperature of Vg" between 

 the two other borders. The slope of temperature in the middle of 

 this plate was probably no more than 10° per linear cM. Hence d 

 was about — G.10~'', and the four (juantities must be calculated 

 from 



. a = 2,98.10-i« i- = — 7,16.10-^ c 1:1.0,132 rf = — 6.10-7 

 Eliminating gp and g» fi'om the equations 1 to 4, we obtain 



v'^ -\- {b -ir (T) V Ar [b d — ac) = . . (5) and u = v -\- {b + d) . . (6) 



or 



uv = ac — b d and u — v ^^ b -\- d. 



Only one of the two sets of values for u and ", furnished by these 

 equations, satisfies all conditions 



« = — 0,005.10-5 1,^7,215.10-^ 



whence we calculate by means of the relations 3 and 4 



.9^=: -0,012 <;„:= 0,133. 



It is evident that gjj and g„ are not even approximately equal. 



The quantities —and — , which according to Riecke's theory are 



proportional to the kinetic energy of positive and negative particles, 

 appear now to be 



.^ = 2,4.10^ i^ = 0,0185.10= 



U V 



2. An objection which may be raised both against this calcu- 

 lation and against that of Riecke is, that the various coeflfieients 

 were not determined all with the same sample of bismuth, and not 

 all in the same magnetic field. Indeed we know how difterent the 

 results of experiments upon bismuth from different sources may 

 prove to be ; and though theoretically all the phenomena are propor- 

 tional to the first power of the strength of the field, practically this 

 proportionality is sometimes far from complete; voN Ettinushausen 

 and Nernst f. i. obtained as constants of the HALL-cttect in mag- 



