126 ' Proceedings of Boyal Society of Edinhurgh. [jan. 16, 
copper wires proceeding to the commutator of a mirror galvanometer. 
These junctions (well varnished) were immersed in a large can of 
cold water, which was frequently stirred. A thermometer was 
placed with its bulb almost touching the hot junction. The heating 
was done by a small spirit-lamp unc'erneath the copper cylinders, 
and the temperature was almost perfectly steady at each measure- 
ment. The resistance of the galvanometer circuit was made very 
large compared with that of the tin and iron which were heated. 
Thus the change of resistance of the glass tube and of the tin in 
melting might be neglected. As will be seen, the cold temperature 
rose slowly ; this small rise of temperature was not taken into 
account. The deflections were each the mean of four (two to each 
side of the scale). 
It was found that up to at least 226° C. the thermometer curve 
was very nearly a parabola. The last column in the table below 
gives the values of the deflection (D), calculated from the formula, 
D = 14-16^ --0207^2 
where ^ = temperature of hot junction - 5°*6 C. As will be seen by 
the table, above 226° C. there is a marked divergence from the cal- 
culated values. Thus it seems that above the melting-point the tin 
line, instead of meeting the iron line at about 342° C. becomes 
almost parallel to the latter. Further investigation, however, at 
higher temperatures, seems desirable. 
Table. 
Temperature C. 
Deflection. 
Cold. 
Hot. 
Observed. 
Calculated. 
5°-6 
00 
o 
952 
952 
5°-7 
119°T 
1846 
1346 
5°-8 
148°*7 
1607 
1602 
6°-0 
l79°-7 
1837 
1837 
6°-l 
226° -5 
2100 
2114 
6°-2 
251°-4 
2162 
2229 
6° -3 
268° -6 
2196 
2292 
6°-4 
285°-8 
2221 
2342 
6°‘5 
297°*8 
2227 
2369 
