602 Proceedings of the Royal Society 
of the areas of these spaces, cdd'c\ is proportional to the electro¬ 
motive force. When the higher temperature, t 3 , is above the neu¬ 
tral point, the electromotive force is the difference of the areas A cc', 
Aee'. The case above mentioned, in which, by a differential 
galvanometer, we get rid of the terms in t 2 , is obviously a process 
for making the curves of two separate complex arrangements into 
parallel straight lines. 
In conclusion, I may give a few instances of the comparison of 
results of calculation of the neutral point of two metals from their 
observed neutral points, and differences of h, as regards iron, with 
calculation of the same neutral point from the portion of the curve 
(assumed to be a parabola) which expresses their electro-motive 
force within ranges of temperature where mercurial thermometers 
can be applied. 
Thus with Fe, Cd, Pb, we have from the iron circuits 0-00112 
- 0*00209 = - 0*00097, while the direct experiment with Cd, Pb 
gave - 0*00096. 
The neutral point, as calculated from the data for the iron 
circuits is - 69° C., while the calculation from direct experiment 
gives - 74° C. 
When the quantities to be found are very small, as for instance 
in the case Ag - Cu, we cannot expect to get a good approximation 
by introducing a third metal. In fact, introducing Fe we find 
indirectly 0*00147 - 0*00151 = - 0*00004, while the direct de¬ 
termination gives - 0*00006. 
Again with Zn and Cu, indirectly we get 
- 0*00042 and - 144° C. 
Directly - 0*00048 and - 146° C. 
Several of the other groups give results as closely agreeing with 
one another as these, others are considerably out. 
The numerical determinations above are founded entirely on a 
series of experiments made for me by Messrs J. Murray and R. M. 
Morrison. Mr W. Durham is at present engaged in determining 
the electromotive force of contact of wires of the same metal at 
different temperatures, with the view of inquiring into its relation to 
ordinary thermo-electric phenomena which appears to be suggested 
by some of the formulas above given. 
