100 REPORTS OF INVESTIGATIONS AND PROJECTS. 



where the temperature changes from one of the end vahies to the other. (In 

 an element used with a cold junction in ice, there are usually two such temper- 

 ature systems in series.) The present paper applies only where this tem- 

 perature distribution exists. In such cases, applying the formula for electro- 

 motive force of a thermoelement which is not perfectly homogeneous, 



-S^ 



( B temperature, H thermoelectric power, £ electromotive force), it is easy 

 lo show that, except for second-order corrections: (i) the temperature meas- 

 ured is that of the junction; (2) but it is measured in terms of the thermo- 

 electric power of that part of the thermoelement which passes through the 

 gradient region of temperature, and not of that lying near the junction; 

 (3) inhomogeneities, permanent or temporary, in the other portions are much 

 less important than in the gradient region. 



This conclusion is little more than an extension of the well-known rule that 

 solder or other foreign material at the junction affects the electromotive force 

 but little. Nevertheless, it seems worth stating, both on account of its impor- 

 tance and because much in the literature seems to point to a very different 

 conclusion and to emphasize the importance of the (usually) really unimpor- 

 tant portion of the thermoelement near the junction. 



The following practical conclusions follow : 



( 1 ) In selecting wire for a thermoelement, the gradient portion is the only 

 part where great care needs to be exercised. The production of a good 

 thermoelement is thus easier than has sometimes appeared. 



(2) If an inhomogeneous element is shifted to a different depth in the bath 

 or furnace, it (in general) at once becomes a different element. The eft'ect of 

 a change in the material near the junction, however, is much less important. 



(3) In cutting off a contaminated platinum thermoelement to restore its 

 original value, it is usually sufficient in practice to cut off only so much that 

 fresh uncontaminated wire occupies the gradient region. 



Commercial constantan wire has been obtained whose original variations in 

 thermoelectric power were o.ooi of the electromotive force against copper. 

 By suitable selection and combination of this wire, thermopiles were easily 

 constructed good to i part in 100,000, that is, 0.001 degree at 100°. These 

 even showed no variation in reading equal to o.ooi degree when the depth of 

 immersion in a constant temperature bath at 100°, and therefore the position 

 of the gradient region, was varied several centimeters. 



(2) Specific heats at high temperatures.* (Abstract.) Walter P. White. Phys. Rev. 

 26, p. 536, 1908. 



The substances under examination were dropped from an electric resistance 

 furnace into the calorimeter described a year ago. The errors of the calori- 

 meter were negligible. The loss of heat in the air was eliminated by first de- 

 termining the heat given out by an empty platinum crucible at various tem- 

 peratures and then subtracting this as an empirical correction from the total 

 heat of crucible and included material. The empty crucible was weighted 

 with platinum so as to keep constant conditions with regard to the splash, etc. 



♦Abstract of a paper presented at the Washington meeting of the Physical Society, 

 April 24-25, 1908. 



