46 Proceedings of the Royal Society of Edinburgh. 
a support, and arranged so that one set of junctions could be pressed against 
one edge of the plate and the other set against the other edge, a very thin 
sheet of mica being placed between the junctions and the plate to prevent 
any direct action of the electric current. 
This sheet of mica, of course, prevents perfect thermal contact between 
the junctions and the plate, and a large correction is necessary on this 
account. The values of the galvanomagnetic temperature effect have 
therefore not the same degree of accuracy as the other transverse effects. 
Nevertheless they are sufficiently accurate to allow interesting conclusions 
to be drawn in connection with the electron theory. 
The thermopile readings were corrected in the following manner : — 
The value of the coefficient S was calculated from readings taken with 
the thermocouples before the thermopile was placed on the plate. It was 
also calculated from observations made with the thermopile. The ratio of 
these two values of S was taken as the correcting factor for the thermopile 
readings in the determination of P. 
Results. 
The values of the four coefficients, R, S, Q, P, for the eight metals tested, 
R 0 R 0 
and the values of the ratios, g, p, p, g, are given in the following tables. 
R x 10 7 . 
S x 10 7 . 
Q x 1 0 4 . 
j P x 10 13 . 
Iron .... 
+ 6-1 
+ 5-2 
+ 9-5 
+ 30 
Nickel 
-33 
-25 
-io-o 
-34 
Cobalt 
+ 2-5 
+ 1-1 
- 7-8 
-22 
Silver 
-3-2 
— 2 - 7 
+ 1-8 
+ 6*3 
Copper 
-2-8 
-2*1 
+ 1-9 
+ 8-2 
Zinc .... 
+ T2 
+ 1*1 
-f- 0*7 3 
+ 4-2 
Cadmium . 
+ 1-2 
+ 0-89 
+ 1-2 
+ 39 
Aluminium 
- i-o 
-0-62 
- 0-42 
- 2-8 
R 
S' 
5 x 10~ 8 . 
5xl0- 3 . 
lb 
Iron .... 
+ 1-2 
+ 3-2 
+ 2-0 
+ 1-8 
Nickel 
+ 1’3 
+ 2-9 
+ 0-97 
+ 4-0 
Cobalt 
+ 2‘2 
+ 3-4 
-IT 
-7*3 
Silver 
+ 1-2 
+ 3-0 
— 5T 
-0-68 
Copper 
+ 1-3 
+ 2-3 
-3 4 
-0-95 
Zinc .... 
+ 1-1 
+ 1-7 
+ 2-9 
+ 0-68 
Cadmium . 
+ P3 
+ 3-0 
+ 3-0 
+ 1-3 
Aluminium 
+ P6 
+ 1-5 
+ 3-6 
+ 0*68 
