6 Messrs. Wright and Thompson on the Determination of 
surround the plate, and probably also with the temperature”. 
In some few cases the constant is of positive sign for some 
solution-strengths or kinds of salts, and negative for others : 
in these cases the numerical value of the thermovoltaic con- 
stant is never great, not exceeding a few centivolts at most. 
If, on the other hand, a metal possess a thermovoltaic constant 
of considerable magnitude for one class of salts, the sign is the 
same for all classes of salts examined. So that practically 
metals may be divided into three classes, viz.:—those where the 
thermovoltaic constant has uniformly a more or less consider- 
able negative value, e. g. lead and silver; those where the 
value is sometimes positive and sometimes negative but never 
large, e. g. copper and cadmium ; and those where the value 
is considerable and always positive, e. g. iron, mercury, mag- 
nesium, and aluminium. 
In order to have a standard for reduction, it is convenient 
to make the same convention as that above referred to for 
voltaic constants, viz. that the thermovoltaic constant for 
amalgamated zinc is taken as zero when immersed in a solution 
of a zine-salt corresponding in nature and molecular strength 
with the solution of the salt surrounding the other plate. 
As regards the actual numerical values assigned below to 
the voltaic and thermovoltaic constants, it has been thought 
best to adhere to the same fundamental assumptions as have 
been hitherto made throughout this series of papers, viz.:—that 
the B.A. unit of resistance is actually one earth-quadrant per 
second; that the factor for converting gramme-degrees into 
volts is 4410; and that the E.M.F. of Clark’s cell at 15°5 is 
1°457 volt, that of a Daniell cell set up with amalgamated- 
zine and electro-copper plates and solutions of zinc and copper 
sulphates of equal molecular strength being °765 x 1-457 
+0005, or 1:114+°0005 volt. The present state of our 
knowledge would introduce corrections in all three of these 
values; but these corrections would so far balance one another 
that the most probable end result is that all the numerical 
values given are about 1°6 to 2°2 per cent. too high. Trirst, 
* §. Czapski has recently found (Annalen der Physik, xxi. p. 209) 
that zinc-silver-chloride and cadmium-silver-chloride cells diminish in 
E.M.F. as the temperature rises; whilst iron-mercurous-chloride, cad- 
mium-mercurous-chloride, and zinc-mereurous-bromide cellsrise in E.M.F. 
with increasing temperature. In the former class of cell the E.M.F. 
actually set up is less, and in the latter gveater, than that corresponding 
with the heat-evolution due to the net chemical change. But it would 
seem that if there be any general law connecting the variations in E.M.F. 
with temperature, it is less simple than might appear from these particular 
results; for Clark’s cells (zinc-mercurous-sulphate) obviously belong to 
the same class as zinc-mercurous-bromide and cadmium-mercurous- 
chloride cells; and yet the E.M.F. of a Clark’s cell falls instead of rising 
as the temperature rises (§§ 139 and 179). | 
