Recalescence of Steel. 5 1 1 



be the considerable difference of temperature evidenced bj 

 reglow, within "05 millim., or we must relinquish the ingenious 

 theory suggested by Prof. Forbes. See also next paragraph. 



III. A rise of temperature takes place throughout the mass. 

 A rod of steel, 1 centim. in diameter and 7 centim. long, was 



softened, and a hole, 1 millim. in diameter and 3 centim. long, 

 was drilled down its axis. A thermoelectric couple of PtCu, 

 composed of wires No. 36 B.W.Gr., was inserted, precautions 

 being taken that there should be contact with the tube only at 

 the junction of Pt and Cu, if at all. The free ends of the 

 couple were attached to a galvanometer which was most satis- 

 factorily dead-beat, a vane dipping in water being attached to 

 the magnet and mirror. The galvanometer-indications were 

 those denoting unmistakable rise of temperature with the 

 reglow during the cooling, and a fall of temperature with the 

 appearance of " darkening " in the heating. 



Considerable difficulty was experienced in showing the 

 actual change of sign in the temperature-variations, until it 

 was realized that care must be taken to apportion the proper 

 masses to the PtCu wires and the steel tube. If the wires 

 were larger than No. 30 B.W.Gr., only a pause in the rise or 

 fall of temperature was observed ; the thermoelectric couple 

 did not truly denote the temperature of the enclosure, the heat 

 coming into the couple being conducted along the thick wires 

 too quickly to admit of a rise of temperature at the actual 

 junction. This was obviated by enclosing the steel tube in a 

 massive brass tube in order to diminish the rate of cooling of 

 the steel, and also by making the thermoelectric couple with 

 finer wire ; the latter plan being found most satisfactory, 

 inasmuch as the reglow could be easily observed. The gal- 

 vanometer-indications denote that the temperature of darken- 

 ing is higher than that of the reglow. 



IV. Peculiarities in thermoelectric properties. 

 Allowances being made for differences in magnitude, the 



peculiarities exhibited by a thermoelectric couple of steel and 

 platinum are similar to those exhibited by the PtCu couple 

 when heated or cooled in a steel tube, as just described. In 

 fact, part certainly of the peculiarities observed and graphi- 

 cally represented by Prof. Tait may be explained by the rise 

 of temperature in the cooling steel, which may, I think, now 

 be taken as proved. 



The same difficulties that I have experienced in proving the 

 rise of temperature inside a block of steel, as a certain point 

 is reached in the cooling of the block, would be met with in 



