92 
PROFESSOR HUGH L. CALLENDAR ON 
condition or temperature, which is the really important point to be considered in 
work of this kind, is increased by the greater complexity and want of symmetry of 
the system of comparison. Examples of the calibration of this box on the binary 
method are given by Barnes, p. 187. 
(19.) Heating of the Thermometers by the Measuring Current. 
It is generally assumed in the construction and use of apparatus for comparing 
standard resistance coils, that the conditions of greatest sensitiveness are obtained 
when the resistance of the ratio arms is equal to that of the resistances to be 
compared. This, however, is certainly not the case in platinum-thermometry, and 
seldom in other cases, unless the heating of the resistances by the measuring current 
can be safely neglected. In platinum-thermometry the heating of the thermometer 
by the current is the limiting consideration which determines the amount of power 
available for the test. For a given rise of temperature with a given thermometer 
the current must not exceed a certain value, e.g ., about a hundredth of an ampere 
for a rise of temperature of a hundredth of a degree, with an average platinum- 
thermometer of •006 // wire (‘Phil. Trans.,’ A, 1887, p. 184). With this limiting 
condition it is easy to see (1) that if the ratio coils are in parallel with the 
thermometer, the sensitiveness is doubled by making the ratio coils very small; 
(2) that if the ratio coils are in series with the thermometer, the sensitiveness is 
doubled by making the ratio coils as large as possible. In every case it is necessary, 
in order to secure accurate compensation for the variation of the leads, that the ratio 
coils should be equal, and that the bridge-wire should be inserted in the circuit 
between the thermometer and the adjustable balancing coils. The arrangement (2) 
with the ratio coils in series with the thermometer, involves leading the battery 
current in through the bridge-wire sliding contact, which is generally unadvisable on 
account of possible disturbance produced by variation of resistance at the contact, or 
by breaking the circuit to readjust. The first arrangement is therefore generally 
adopted, and some advantage is gained in this case by making the resistance of the 
ratio coils considerably smaller than that of the thermometers, provided that they 
are not made so small as to be appreciably heated when the working current is 
passed through the thermometer. In this particular box the ratio coils had a 
resistance of 6’4 ohms, or about a quarter of the normal thermometer at 0° C. 
They were each constructed of two '008" platinum-silver wires in parallel, and 
adjusted for equality of temperature-coefficient. With a working current of '002 
ampere through a 25'6 ohm. thermometer constructed of '004" wire, the heating 
effect in the thermometer would be nearly a thousandth of a degree, and the current 
through the ratio coils would be nearly one-hundredth of an ampere, which would 
not produce any material rise of temperature. 
