462 Mr. F. W. Jordan on tlie Direct 



divided the lump along a plane inclined to the axis of figure 

 of the crystallized mass. The brilliant surfaces of cleavage 

 could be traced by chipping and cutting to the middle of the 

 lump, and were found to be nearly plane and parallel to 

 each other. The rod of bismuth in this experiment was cut 

 from the middle of the crystallized mass so that its axis was 

 parallel to the principal cleavage plane of the bismuth. The 

 surfaces of contact of the bismuth rod with the copper were 

 at right angles to the principal cleavage planes. 



This method of preparing' the specimen of crystallized bis- 

 muth is, with the exception of the electrical heating, similar to 

 that employed by Perrot *. He prepared several crystals from 

 the same mass of bismuth and measured for each the thermo- 

 electric power with copper in two different positions. In 

 one of these positions, the surfaces of contact with the copper 

 were parallel to the principal cleavage plane of the crystal, 

 and in the other position they were at right angles to this 

 plane. The first of these positions was designated by the 

 symbol II and the second by the symbol ±. He found that 

 the thermoelectric power in the position II was approximately 

 twice as great as in position j_ at a temperature of 55° 0. 

 The thermoelectric powers at a temperature of 55° C. for 

 four of his crystals in position j_ are given here for the 

 purpose of comparison with the value derived from the 

 Peltier coefficients in the table above. 



Thermoelectric Power at 55° C. 

 Crystal P ± ... 53*4 X 10~ 6 volt per degree. 



„ (j J. ... 51*1 ,, „ ,, 



55 ■«■ J- ... 0O*O ,, ,, 55 



„ M± ... 55*5 „ 



The lengths of these crystals varied from 19 mm. to 

 30 mm. In the table above the thermoelectric power as 

 derived from the Peltier coefficient would at 55° 0. be equal 

 to 57*6 X 10~ 6 volt per degree. Considering the widely 

 different results that Perrot obtained, this value is quite 

 acceptable. The bismuth used by Perrot was analysed by 

 three chemists and found to be pure, with the exception of an 

 undetermined trace of iron. The different results obtained 

 by him are probably due to slight irregularities in the 

 structure of the crystals. The bismuth rod in this apparatus 

 has not yet been analysed, and consequently no accurate 

 comparison of the results can be made. 



* F. L. Perrot, Arch, des Sciences Phys. et Nat. Aug. 1898. 



