392 Proceedings of Royal Society of Edinburgh. [july 18, 
German silver strips had to be brazed to the nickel, that the 
junctions might stand the high temperatures. The neutral point 
of the pair was found as usual by long strips cut from the same 
sheets. Since, in the brazing, the metals had to be brought to a 
high temperature, these long strips, as well as the pieces that were 
to be brazed, were all annealed by heating to bright redness and 
slowly cooling. 
In order to measure the thermoelectric power of nickel German 
silver at the various temperatures, the following method was 
used : — The current, instead of being sent through Ni-Arg junc- 
tions, was sent through the measuring (FeArg) thermopile, and the 
Peltier effects caused in the thermopile junctions measured by 
connecting the nickel and German silver strip to the galvanometer. 
The deflection here in the galvanometer would be proportional to 
the product of the Peltier effect of FeArg, and the thermoelectric 
power of NiArg. Now, it has been shown by former experiments 
by the writer* that the Peltier effect of FeArg varies as the 
absolute temperature. Hence we can at once find the thermo- 
electric power of the NiArg at any temperature. 
The junctions in this case were packed in four thick copper 
tubes, one inside the other, with asbestos wool between. As this 
arrangement gave a very uniform temperature at the junctions, most 
of the readings were taken with the temperature rising slowly, 
except those at 17°*8, 19 o, 0, 23 o, 0, and 254 0, 3 C., when the tempera- 
ture was fairly steady. The temperatures above 285° C. are not 
very certain. For convenience in heating, the tubes had to be 
almost horizontal. This, unfortunately, caused the mercury in the 
thermometer to boil at a temperature much below its boiling point 
at atmospheric pressure. The readings 340° and 330°, therefore, 
are only estimated. 
In Table IV. are given some of the measurements of the Peltier 
effect in NiArg. The second column gives D/Ctf, where D is the 
FeArg thermopile deflection, C the battery current through NiArg, 
and t the absolute temperature. In Table V. are the measurements 
of the thermoelectric power of NiArg (by sending the current 
through the FeArg thermopile). The second column gives the 
* Proc. Roy. Soc. Edin ., 1882. If we introduce the specific-heat correction 
(12 % per ° C.) we get a much nearer agreement than that shown there. 
