416 M. Regnault on the Measurement of Temperatures 

 Second Series. Third Series. Fourth Series. 



I have represented graphically the results of these experi- 

 ments. To do this easily I took for ordinates the differences 

 of temperature of the iron and platinum junctions divided by 

 3, and I took for abscissae the differences of temperature of 

 the bismuth and antimony junctions multiplied by 5. 



The conditions being apparently identical in these several 

 series of experiments, the corresponding curves ought to be 

 susceptible of being superposed : but such is not the case ; in 

 some instances the curve exhibits a very satisfactory regularity 

 throughout its extent; in other cases, on the contrary, and 

 without its being possible to ascertain the cause, there is a 

 sudden leap at a point, and the second portion of the curve 

 no longer agrees with the first : very rarely the curves fur- 

 nished by two series of experiments approach sufficiently for 

 the differences to be attributed to errors of observation, and 

 the two curves to be considered as the expression of the same 

 phaenomenon. 



These variations are probably owing to changes which take 

 place in the molecular condition of the metals at the place of 

 the solderings, and which suffice to modify perceptibly the 

 electromotive forces. Sometimes these changes come on sud- 



