376 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



The h curve was determined by observations on the same section as in 

 the experiments vrith 35 amperes. Referring to Fig. ^,fg is the curve ob- 

 tained ; it is quite straight and agrees with the h curve of Fig. 3 very well, 

 as far as it goes. The bend in the latter depends principally on the obser- 

 vation at 35 amperes, which was, of course, not taken in this experiment. 



The results of this experiment are of the same general character as in 

 the experiment with 35 amperes. There is as before a decrease in the 

 value of the Thomson Effect for higher temperatures, with the exception 

 of the shortest section, that is, 17 to 13. This is due to the very flat 

 top on the curve c d, Fig. 5. The curve c d in this experiment is not 

 nearly so good as in the experiment with 35 amperes. This is unfor- 

 tunate, as the whole calculation for the Thomson Effect with 30 amperes 

 depends on this curve. 



For this reason, another determination of the change of temperature, 

 on reversal of the heating current, was made for the several sections, 

 again using 30 amperes. More observations than before were taken, so 

 that it is hoped the means are more accurate. Great care in every 

 respect was taken. The resulting curve is shown dotted in Fig. 5. Its 

 general form is more similar to the curve c d for 35 amperes, than the 

 previous one, although the values of its ordinates do not differ greatly 

 from it, the curves crossing each other at several points. 



The principal quantities required for the calculation of the Thomson 

 Effect as determined from this new curve are given below. 



