376 BRIDGMAN. 



loaded, the material of the two branches of the couple now becomes 

 effectively different, and a thermal e.m.f. appears. This was meas- 

 ured as a function of tension and temperature difference. 



\^arious precautions were taken in performing the experiment 

 which need not be gone into here at length. The two wires passed 

 through a heavy glass tube for their entire length, and in this way 

 were protected from contact with the water of the bath. To secure 

 temperature equality and diminish error from heat conduction along 

 the wires from one bath to the other, the glass tube was filled, up to 

 the level of the water in the upper bath, with Bureau of Standards 

 resistance oil. As a further precaution against slight variations in 

 temperature, the glass tube was covered for the entire length of 

 immersion in the upper bath with a compound tube of \ inch thick 

 copper on the inside fitting closely into an outer iron tube also | inch 

 thick. Since the lower bath was always filled with ice and conse- 

 quently was not exposed to slight oscillations of temperature, such a 

 precaution was not necessary below. The approximate dimensions 

 were; depth of each bath 1 ft., free distance between baths 2 ft., 

 diameter of glass tubing 1 inch. 



It has been already stated that the results of these experiments 

 were very irregular, the eflfect varying with the nature of the speci- 

 men. ,In the following the nature of the results will be briefly indi- 

 cated; each result given is the mean of determinations on the right 

 and the left hand wires. 



Nickel. Wire from the same manufacturers but from a dift'erent 

 spool from the pressure sample was used. This was 0.005 inches in 

 diameter. A maximum tension of 2000 kg./cm.^ was applied. The 

 results were fairly good; the two samples agreed on the average 

 within about 5%. The thermal e.m.f. may be represented by the 

 following equations : 



At 31° e.m.f. = (.OOSSST-O.OeGOT^) X 10~« volts, 



at 51° = (.00577T-0.05132T2) X 10-«, 



at 77.5° = (.0081 T-O.OsHST^) X 10-«, 



at 94.5° = (0121 T-O.O537T2) X IQ-^. 



In these formulas T is the tension in kg./cm.^ The direction of 

 the e.m.f. is from unstretcheH to stretched at the hot junction. Paying 

 attention only to the change of volume, this would correspond to a 

 pressure effect of from compressed to uncompressed at the hot junction 

 This is of the opposite sign from the observed pure pressure effect. 



Copper. This was commercial wire of 0.020 inches diameter. 



