88 
PROFESSOR HUGH L. CALLEXDAR OX 
annealed after winding, and that its resistance is consequently liable to change for 
some time. (2) That the temperature-coefficients of different specimens are often 
different, and vary for different sizes according to the method of annealing, so that it 
is desirable to test the temperature-coefficient of each coil in the box, and to apply 
the corrections separately in the most accurate work. (3) That the temperature 
correction cannot be satisfactorily applied, as there is generally some hysteresis in the 
change of resistance with temperature, and the values of the resistances depend to 
some extent on previously existing conditions of temperature. (4) That it cannot he 
hard soldered without burning, and that soft solder connections to manganin are 
frequently found to be defective, unless they have been most carefully made. 
(5) That it is liable to corrosion if exposed to damp or gas fumes, although the usual 
coating of shellac is sufficient protection in most cases. These defects have been 
noted by other observers, notably by Harker and Chappuis, who employed a box 
with manganin coils in their recent comparison of the platinum and gas-thermometers 
The majority of other alloys of this class are inferior to manganin in constancy; they 
are also frequently objectionable (e.p., “ constantan”) on account of their great 
thermoelectric power, which produces inconvenient disturbances if the temperature ot 
the box is not uniform. The advantage of platinum-silver lies in the perfection with 
which it may be annealed, and in the absence of lag or change when properly 
annealed. 
If uncompensated coils of platinum-silver are used, it is necessary to keep them in 
an oil-bath to secure sufficient certainty of temperature. It is also desirable to 
employ a thermostat for regulating the temperature of the oil-bath, and a stirrer for 
keeping the temperature uniform throughout. This adds considerably to the cumber¬ 
someness and expense of the apparatus, and to the difficulty of using it. It is really 
simpler in the end to use compensated coils, as the individual temperature-coefficients 
of the coils must otherwise be determined and corrected separately, at least for work 
of the highest accuracy. 
The first apparatus constructed on this principle was made in 1887, shortly before 
applying for the patent, and was figured in the specification. All the parts of this 
apparatus were made interchangeable in pairs, by an extension of the Carey-Foster 
principle, with a view to facilitate testing and calibration. There w,ere two exactly 
similar bridge-wires, each a metre long, and three pairs of compensated coils, which 
could be inserted singly or in series, with resistances on the binary scale equal to 1,2, 
and 4 times that of either bridge-wire. The platinum-thermometers were also made 
in pairs, after the pattern described and figured in the ‘ Phil. Trans.,’ A, 1890, but 
were generally contained in separate tubes, which greatly facilitated construction, 
and permitted them to be used differentially by a simple change in the connections of 
the bridge. In the instrument used for the boiling-point of sulphur ( loc. cit .) it was 
necessary to have both thermometers in the same tube. This construction was also 
adopted in 1887 for comparison coils in which it was necessary that both wires should 
