( o.>.) ) 



of platinum acted quite differently in this regard. The strongest 

 thernio-currents were obtained witii Pt Jr of 10 to 20 7o- ^ yv'ivQ 

 of this platinum was fused into each of the legs of the f/-tube, as 

 accurately as possible in the middle-section. These wires being con- 

 nected and a sensitive galvanometer being introduced into the cii'cuit, 

 the temperature-differences ^Thu and A ti could be measured in 



proportion. Should we have wished to measure each of tliose (pianti- 

 ties separately', it would have been necessary to determine the thermo- 

 electric constants of this platinum with regard to mercury. 



The unequalify in temperature in the middle wires caused by an 

 inevitable lack of symmetry in the L-^-tube was compensated by 

 means of another thermo-couple. After each series of observations 

 the galvanometer deflection, given by this couple with a known 

 resistance and a known difference in temperature between the points 

 of contact, was measured, in order to eliminate changes in the sensi- 

 bility of the galvanometer or in the distance of the scale. 

 jj 



The quotient — was determined indirectly. If the external exchange 



of heat could be neglected, the temperature-gradient must be the 



same all over the parts of non- uniform temperature, so long as the 



current did not pass through the mercur)-, apart from the distribution of 



temperature near the limits. And in the middle-section the gradient 



of temperature would remain very approximately the same, when the 



U 

 current did pass through it. Therefore the quantity — coukl be said to 



be equal to the temperature-gradient in the middle-sections. 



To measure this gradient in each of the legs of the f/-tube on 



both sides of the middle-section at a given short distance both above 



and below it, another wire of platinum was fused in. The tempe- 



raturerdilference between these sets of wires divided by their distances 



U 

 Avas put for —. 



The wires last mentioned were of a kind of platinum of which 

 the thermo-electrical constants with regard to mercury had been 

 accurately determined beforehand. As the same thing cannot be said 

 about the wires in the middle-sections it is impossible to .say any- 

 thing definite about the uniformity of the gradient resulting from the 

 experiments as they have been made. Preparatory experiments 

 however have showji, that when / does not exceed certain limits, 

 the gradient is sufïïciently uniform. 



Much trouble has been caused bv wild thermo-electric currents. 



