168 INSTRUMENTATION IN SCIENTIFIC RESEARCH [Chap. 2 



thermal emf is in general several millivolts for a temperature 

 difference of 100°C. 



The emf is usually measured by cutting one of the conductors, for 

 instance at the point M , and inserting a meter. Since the connecting 

 wires are in general made from a material C, other than A and B, the 

 circuit assumes the form of Fig. (2-1)12. The sum total of all emfs 

 in such a circuit is zero as long as all connecting points P 1 , P 2 , and P 3 

 are at the same temperature t v However, if the point P 2 is exposed 



— -j , to a temperature t 2 =t x — At, an emf 



c ' f*\ W *H arise in the circuit, which is again 



c yJ to a first approximation proportional 



— ^ ' to At. The magnitude of the thermal 



emf is independent of the presence of 



P 3 I*-*, 



H l^ 



\ t the third metal C as long as the contact 



- ^"IA u-b~ points P x and P 3 are at the same tem- 



r y "1 perature. If the temperature of P x 



and P 3 is not the same, the total emf 



developed in the circuit is equal to the 



Fig. (2-1)12. Thermoelement algebraic sum of the thermal emfs de- 

 made from two wires A and B; i i , v • ,- 



. „ veloped at each l unction. 



junction point /, at temperature r J 



t 2 ; the wires A and B are con- The thermal emf is usually measured 



nected, with a wire c made from either with a galvanometer (millivolt- 



a third material, to the meter M. i _\ ±4-- „4.„ i,j.i c 



„. D , D i meter) or a potentiometer, both ol 



The junctions I J l and i J 3 are kept ' L 



at the temperature t v which can be directly calibrated in 



terms of temperature. The galvanom- 

 eter measures the emf indirectly by measuring the current 

 produced by the emf. Since the current depends also upon the 

 resistance in the circuit, the measurement is correct only if the 

 resistance is constant. Variation of temperature of the wires tends 

 to vary the resistance, thus introducing an error in the measurement. 

 The error can be minimized if the resistance in the circuit is large 

 compared to such resistance variations. Since it is desirable that 

 the greatest voltage drop is across the meter, meters with relatively 

 high resistances are employed, although the use of such meters 

 reduces the sensitivity of the measuring system. 



A current in the thermocouple circuit tends to develop heat in the 

 wires as well as in both junctions (Joule effect). A further influence 

 of a current in the circuit is a temperature increase of one junction 

 and a temperature decrease of the other junction (Peltier effect). The 

 magnitude of these effects is frequently negligible. The current 

 errors can be completely avoided by measuring the thermal emf with 

 a potentiometer method. 



