162 
MR. H. L. CALLENDAR ON THE PRACTICAL 
lias been abundantly verified in all the experiments undertaken with it. If the 
wire is pure to start with, and care is taken not to alloy it with silicon, carbon, tin, or 
other impurities, and not to subject it to strain (such as increase of length), its 
resistance is always the same at the same temperature ; at least, this was the case 
with the specimens of platinum wire used in these experiments. For the proof of 
these assertions the reader is referred to the experimental details (especially pp. 
192, 205 et passim). 
The wire used here is obtained from the well-known firm, Messrs. Johnson, 
Matthey, and Co., and is probably very pure and well adapted for the purpose 
(see p. 182). 
Platinum wire evidently fulfils the conditions laid down in II., since it is essentially 
a standard of electrical resistance. 
The method of comparing and verifying copies of the standard is of fundamental 
importance and requires a more detailed explanation. The great advantage of the 
proposed platinum standard over all others is the ease and accuracy with which this 
may be effected, a degree of accuracy greatly exceeding that of all other thermo¬ 
metric comparisons, and almost independent of the distribution of temperature. 
The wires are wound side by side on non-conducting material and symmetrically 
disposed, so that their mean temperatures are always the same and simultaneous 
values of their resistances are observed. Observations are also taken in melting ice 
and in steam at atmospheric pressure for the purpose of verification and to test for 
changes of zero. Details of the method and of the experimental comparison of wires 
of different metals are given in a subsequent section. An accuracy of 1 in 10,000 is 
attainable. The general conclusions from these experiments are as follows :— 
[If If be the resistance of a given wire at the temperature t Cent., and II 1; B 0 , 
the values of its resistance at 100 ° and 0 ° C. respectively, (E^/Rq — l) is the increase 
of resistance between 0 ° and 100 ° C., and is found to vary considerably for different 
specimens of commercial wire of the same metal. ( E.g ., for platinum, it varies from 
0 - 25 to 0‘35.) But the curves of resistance variation are similar, so that the values of 
the function (R/R 0 — 1 )/(R 1 /R 0 — 1 ) are nearly identical for different wires of the 
same metal through a range of 600° C., though they differ very widely for different 
metals. The variations are mainly due to slight chemical impurities, which suffice 
materially to reduce the temperature-coefficient and make the wire much less suited 
for thermometry.] 
Pure platinum wire, when once annealed, is little liable to alteration. If it is 
pure, different lengths from the same bobbin do not behave differently, as some 
observers, perhaps using less pure wire, have stated. Its resistance is not liable to 
permanent change by mere heating and cooling, provided the wire is not strained or 
chemically altered. Different specimens of pure platinum wire agree very approxi¬ 
mately throughout the scale. This is very convenient, but not essential; for, if they 
did not, they could be accurately compared and the differences tabulated. It 
