( 1-^) 



netic fields from 1650 to 11100 values from — 10,27 to — 4,95. 

 Moreover in the formulae the value of the resistance of bismuth 

 must be used, which depends largely on the strength of the mag- 

 netic field. These considerations induced me to repeat the measure- 

 ments of the various phenomena with the same plate of bismuth 

 and in the same magnetic field i). 



In order to avoid as much as possible the disturbances which 

 might be caused in cast plates by irregularities in the cristalline 

 structure ^), these experiments were made with a plate of bismuth 

 deposed by electrolysis in a solution of nitrate of bismuth. 



During the experiments the plate was fastened on a little wooden 

 board under two copper strips, soldered to the copper tubes ser- 

 ving to supply or absorb heat and at the same time as electro- 

 des for the galvanic current through the plate. The thermo-electric 

 needles could be introduced between the plate and the wood at three 

 different, fixed distances from the copper plates, for which purpose 

 three grooves had been sawn in the wood. In measuring differences 

 of potential the copper wires of the thermo-electric couples (German 

 silver-copper) were used. The space between the plate and one of 

 the poles of the magnet was filled up with cotton. 



A similar arrangement of the experiments was used by voN 

 Ettingshatjsen and Nernst in their research on the phenomena in 

 pure bismuth and in alloys of bismuth and tin ^), with this diffe- 

 rence, that as well the copper tubes as the thermo-electric needles 

 were soldered by them to the plate. 



3. Measurements. Dimensions of the plate: length 8,8 cM., 

 breadth 1,15 cM., thickness 0,089 cM. 



The experiments in which the thermo-electric needles were on the 

 middle of the plate are indicated by the letter A ; those in wliich 

 they were placed on the side of the heated tube by B and the third 

 series by C. The distance between the two copper strips was 2,96 

 cM; the distances between the heated strip and the 1^*, 2'^ and 8'^ 

 groove were respectively 0,65, 1,53 and 2,37 cM. 



In judging the results, entered under B and C we should keep 



1) See my communication in the Meeting of May 28, 1898, p. .53. Communications 

 Phys. Lab. Leiden, N". ■11, p. 13. 



-) See my communication in the Meeting of April 21, 1897, p. 501 and June 215, 

 1897, p. 71, 72. Comm. N". 37, p. 17, N". 40, p. S, 9. 



'■') Wied. Ann. 33, p. 474. If in the course of this research also the rotation of 

 the isothermal lines had been measured, it would have furnished good data for a 

 calculation. 



