600 PROFESSOR MACGREGOR AND DR KNOTT ON 
soldered to thick copper wires, however, there was danger that the heat would 
thus be conducted rapidly away, and their ends thereby rendered cooler than 
their middle parts. To prevent this the alloy was always sunk about five 
inches beneath the surface of the oil, and the oil kept for a considerable time 
at approximately the temperature at which the resistance was to be measured. 
With these precautions, we-considered ourselves justified in assuming that the 
lower ends of the thick copper wires had the same temperature as the oil at 
the moment of measuring the resistance. 
That we might be sure that the thermometer had the same temperature 
as the oil, it was necessary (as it was not possible to keep the oil for very 
long times at permanent temperatures) to use a thermometer with as small a 
bulb as sufficient accuracy in readings would allow. We chose one which 
would easily give accurate readings to one-tenth of a degree Centigrade. 
We determined its errors between the limits of temperature for which we 
used it by comparing it carefully with another thermometer whose errors 
had been determined at Kew. Our temperature limits were about 16° C. 
and 150° C. 
The copper connections were so thick that their resistance and the varia- 
tion of the resistance of those parts of them which were heated could be 
neglected. We satisfied ourselves by special experiments, the details of which 
need not be given, that a thin lamina of the oil which we used had a resistance 
greater than we could measure with the coils at our disposal, and a conductivity 
therefore which might be neglected. 
Having only very small quantities of the alloys we could not determine 
their coefficients of expansion. We are therefore unable to eliminate the 
change of resistance due to the change of dimensions of the wires. If we 
assume the mean coefficient of expansion per degree between 0° and 150° to 
be 00002, then a wire whose resistance at 0° was unity would, through change 
of dimensions alone, be only about ‘997 at 150°. For all alloys whose co- 
efficients of expansion are positive the increase in the specific resistance of the 
alloy is always greater than the increase of the resistance of the wire which is 
measured. But without the coefficient of expansion it cannot be determined. 
In this paper, then, we discuss the variation of the resistance of wires of certain 
alloys. 
Iron-Gold Alloy. 
Table I. gives the results of experiments with the only alloy of iron and gold 
which we had at our disposal. It contained 5 per cent. by mass of iron, or 
approximately 11‘9 per cent. by volume. The first column gives successive 
temperatures, the second the corresponding observed resistances of the wire. 
