360 Dr. 0. Barus on the Relation between the Thermoelectric 



this constant for glass-hard rods is remarkably different. The 

 rods in Table VI., for instance, possess a T. E. H. amounting 

 nearly to 140 : 10' 5 ; whereas in the rods in Table II. the maxi- 

 mum * value found does not exceed 70 : 10 6 . This may be 

 due to a difference in thickness, or, more probably, to a dif- 

 ference in the composition of the rods examined. 



These phenomena were further studied through the follow- 

 ing experiments. 



1. Commercial rods of different diameters were glasshard- 

 ened and examined with reference to the current produced 

 when one end of a couple was cooled with a wedge-shaped 

 piece of ice. In general, a maximum of T. E, H. was ob- 

 served in rods whose diameters lay between 1 and 2 millims. 

 These experiments, however, are unsatisfactory, inasmuch as 

 the composition of the rods enters as an element of disturb- 

 ance which cannot be allowed for. For this reason experi- 

 ments were made on thick bars, the parts of which had been 

 filed to different diameters. 



2. The halves of each of two pieces cut from the same 

 rod 5 millims. in diameter were reduced f by filing to thick- 

 nesses of 3 and 1 millim. respectively, and glasshardened. 

 During the process of heating, care was taken to raise all parts 

 of the bars to the same degree of redness. On connecting 

 the ends with the galvanometer and applying the ice wedge 

 at the middle (where the diameter enlarged), very decided 

 currents were observed passing from thin to thick through 

 warm. Hereupon two cones were filed from the same mate- 

 rial (5 millims. base and 50 millims. long). Point and base 

 of the cones (previously glasshardened) being connected with 

 the galvanoscope, the ice wedge applied at any point pro- 

 duced in each case currents from apex to base through warm, 

 thus harmonizing with the previous experiments. Near the 

 points only the results became uncertain. On bringing toge- 

 ther the cones with the rod [IV] (Table VI.), the points were 

 found thermoelectrically harder, the bases softer, than the 

 former. On the other hand, the point of a fine needle pre- 

 pared from the same material gave contrary indications ; the 

 fine point therefore was apparently softer than rod [IV]. 



3. Finally, two very gradually tapering cones were pre- 

 pared from another steel rod 2*8 millims. in diameter. Con- 



* These rods (Table II.), even when heated to the utmost white and 

 suddenly cooled, remained strongly electropositive towards copper. 

 T. E. II. therefore even in this extreme case was much less than 107 : 10 6 . 



t It is to he observed that the thinner parts of these pieces sooner 

 arrive at red heat and remain longer in this condition than the thicker 

 parts. This applies equally to the cones. 



