55 



brought back till the galvanometer shows no current ; and then (by 

 a process quite analogous to that followed by Mr. Joule and Dr. 

 Lyon Playfair in ascertaining the temperature at which water is of 

 maximum density) the temperatures of the two junctions are ap- 

 proximated, the galvanometer always being kept as near zero as 

 possible. When the difference between any two temperatures on 

 each side of the neutral point which give no current is not very great, 

 their arithmetical mean will be the neutral temperature. A regular 

 deviation of the mean temperature from the true neutral tempera- 

 ture is to be looked for with wide ranges, and a determination of it 

 would show the law according to which the difference of the spe- 

 cific heat of electricity in the two metals varies with the tempera- 

 tures ; but I have not even as yet ascertained with certainty the ex- 

 istence of such a deviation in any particular case. The following is 

 a summary of the principal results I have already obtained in this 

 department of the subject. 



The metals tried being, three platinum wires (P l the thickest, 

 P 2 the thinnest, and P 3 one of intermediate thickness), brass wires 

 (B), a lead wire (L'), slips of sheet lead (L), copper wires (C), and 

 iron wire (I), I find that the specimens experimented on stand 

 thermo- electrically at different temperatures in the order shown 

 in the following Table, and explained in the heading by reference to 

 bismuth and antimony, or to the terms " negative " and " positive " 

 as often used : 



Temp. 

 Cent. 



Bismuth 

 Negative." 



Antimony 

 " Positive." 



-20 







37 



64 



130 



140 



280 



300 



P P 



r 2 r i' 



...Po. ...C P,. 



.P 3 *...{KP,} C...P! 



P 3 P 2 ... ...*_ 



.p, p a .. 



.1. 



C I. 



...C ...I. 



b I C 



