630 
Proceedings of the Royal Society 
curve obtained (see diagram, Fig. II.) being throughout the same 
range of temperature very accurately a straight line. The resistance 
was measured by shunting the current in the galvanometer and 
battery circuit through the wire under consideration. One great 
disadvantage of this method is, that the curves do not represent 
strictly the relation between temperature and resistance, since the 
rate of cooling is not uniform, and the wire is not at one tem- 
perature throughout. 
In the second method, however, this difficulty was overcome ; 
for time, as a variable, was eliminated by combining the two ori- 
ginally separate experiments with iron and platinum into one, and 
comparing the simultaneous changes in the resistances of these 
wires, which were in exactly similar circumstances. Equal lengths 
of iron and platinum wire were led side by side through the hori- 
zontal cylinder, and their extremities were so connected with the 
galvanometer and battery circuit, that by simply rocking a six- 
footed rocker, working in six mercury holes, the current could be 
shunted through each wire alternately, and thus their resistances 
could be compared by the effects produced upon the galvanometer. 
The curves obtained from these experiments with platinum and 
iron (see diagram, Fig. III.), their indications being here abscissse 
and ordinates, show the same marked change at the same critical 
temperature. When palladium was substituted for platinum the 
same peculiarity was observable (see diagram, Fig. IV.); but when 
palladium was substituted for iron, the curve obtained (see diagram, 
Fig. V.) was an accurate straight line. It was found expedient, 
after the first few. preliminary experiments, to introduce into the 
battery circuit a commutator, by which to reverse the current, and 
so eliminate all errors referable to thermo-electric effects due to the 
unequal heating (by radiation from the cylinder, or conduction along 
the heated wires), of the various metallic junctions in the circuit. 
In the third distinct series of experiments the arrangement was 
more elaborate. To the platinum or other wire, whose resistance 
was compared with that of iron, was attached at the middle point 
a third wire. By a mechanical arrangement of rockers and com- 
mutators, the battery and iron wire could be thrown out of the 
galvanometer circuit, and thus a thermo-electric standard of 
temperature was obtained, with which the resistances of the two 
