226 
MR. H. L. CALLENDAR ON THE PRACTICAL 
amounting to nearly 1° C. was caused in each wire by the half-hour’s exposure to the 
temperature of 800° in contact with clay and in the presence of copper, and probably 
tin and zinc. The comparison at higher temperatures has to be effected by a different 
method to secure good insulation and guard the wire from contamination, but the 
experiments have not yet been completed. The results of the comparison are shown 
graphically in Plate 13, fig. 10. The dotted portions of the curves are somewhat hypo¬ 
thetical, but there is naturally a family resemblance between them. They all show a 
comparatively sudden change (due to ph) somewhere between 200° and 300°, which, 
though small in amount, is distinctly marked; it is in each case followed by a 
flattening due to similar changes in the other wires, differing chiefly in phase. A 
similar deviation is also indicated by the most accurate of the air thermometer obser¬ 
vations (see page 221). The differences between different thermometers of pure 
platinum wire are seen to be very small, and to be capable of accurate measurement. 
I do not think they can possibly he due to any constant error affecting the method of 
comparison, as I have been unable to think of any that would at all account for the 
observed phenomena ; but I have procured a stock of various specimens of platinum 
for more extended investigations. 
Comparison of Platinum and Iron . 
The comparisons of platinum and iron wires given in the earlier part of the paper 
were felt to require confirmation. The agreement with the exponential empirical 
formula was so close that I was inclined at the time to think that the deviations from 
it were errors of observation without meaning. Commercial tinned iron wire had 
been used (the tin having been removed with glass paper as far as possible), and 
sufficient precautions had not been taken to avoid oxidation. The wire had ulti¬ 
mately reached a very steady state, but not until it had become covered with a film of 
black oxide. A new comparison coil was therefore wound with pure iron wire and 
platinum wire No. 2. By an improved method of winding very good insulation was 
secured, and the wire was carefully heated in dry nitrogen and very little oxidised. 
(The steam which always escapes from the clay when heated probably produced the 
slight oxidation observed.) The results of this comparison are given in Table F.-n. ; 
they are simultaneous mean values of the resistances deduced, as before, from several 
observations at steady temperatures. 
To facilitate the comparison of the two series of observations, the results of the 
first series are reproduced in Table F.-l, and reduced on the same plan as those in 
Table F.-ii. 
