418 M. Regnault on the Measurement of Temperatures 

 Seventh Series. 



I represented by the graphic method the results of these 

 new experiments as I had done for those of the first. I ascer- 

 tained thus that the three curves could not be superposed, but 

 that nevertheless they exhibited less considerable deviations 

 than those of the first four series. 



In the second series the oil-bath was allowed to cool after 

 the observation made at 281°, without disturbing any portion 

 of the apparatus, and the experiments recommenced when 

 the temperature of the bath had descended to 140°. It should 

 be observed, that the portion of the curve corresponding to 

 this second period does not agree with that which is given of 

 the first, and nevertheless the apparatus had undergone no 

 change. 



I have made numerous experiments on the bismuth and 

 antimony couple in order to ascertain whether these irregu- 

 larities were not principally due to that couple ; but by vary- 

 ing the temperature of the junctions between the limits which 

 had been attained in the preceding experiments, viz. from 15° 

 to 33°, I found that the bismuth and antimony element re- 

 mained pretty constant. I observed that a difference of tem- 

 perature of 1° between the two solderings produced sensibly 

 the same deflexion of 17° on my galvanometer, whatever was 

 their absolute temperature, provided it always remained com- 

 prised between the limits above indicated. But it is difficult 

 to decide whether this proposition is exact, or merely approxi- 

 mate, because the intensity of the current varies perceptibly 

 with the time, even when the two junctions constantly present 

 the same difference of temperature; and there always remains 

 a little uncertainty as to the value of the deflexion which 

 should be inscribed. 



But I found, contrary to the opinion generally admitted, 

 that an increase pf 1° in the difference of temperature of the 



