AS DETERMINED BY FIVE PLATINUM-RESISTANCE THERMOMETERS, 245 
A complete determination of temperature on the platinum scale by means of one 
of the sunken thermometers is, therefore, reduced to the following simple steps :— 
(1) The balancing of the galvanometer and reading of the bridge wire scale (R) 
and the temperature (0) of the box. 
(2) To It is to be added the correction for the particular arrangement of coils used, 
from Table I. 
(3) The correction to reduce the bridge wire reading to mean box units, from 
Table II. 
(4) The reduction to standard temperature (14°). The quantity taken from 
Table III. multiplied by (6 —14) gives this correction. 
(5) The correction from Table IV. 
It only remains to reduce the temperature thus expressed from the platinum to 
the air scale. 
The relation connecting these two, established by Professor Callendar, # is 
in which pt is the platinum temperature, t the temperature on the air scale, and S 
a constant. 
For a completely independent standardisation it would be necessary to determine 
the resistance at some third known temperature in order to obtain the value of §, 
but the experiments of Callendar and Griffiths have shown that although the 
value of S varies from one specimen of platinum to another, it is a constant for any 
particular sample of wire. References to the original papers bearing on this point 
are given in Mr. Griffiths’ article in ‘ Nature ’ cited above. 
The value of S for the particular wire used in the Oxford instrument was deter¬ 
mined at Cambridge to be P512.f If it were intended to employ the Oxford 
apparatus for the determination of temperatures over a very wide range, it would 
doubtless have been advisable to make an independent determination of the value of 
this constant. Since, however, the range —15° C. to +25° C. will cover all the 
variations of earth temperatures with which alone we are here concerned, and since 
within that range the correction does not amount to as much as 0‘3, an error of even 
0’050 (which is quite inadmissible) in the value of § would not affect our results. 
Writing pt -f- d for t in equation ( h ), and remarking that since c//100 is less than 
(BOOS, its square may be neglected, we find 
d — S(t 2 —■ t)/{ 1 -f- (l — 2r)S/100} 
t being written for jiZ/100, 
* ‘Phil. Trans.,’ A, 1887. 
t Cf. the Report of the Committee of the British Association for improving the Construction of 
Practical Standards for use in Electrical Measurements. Bradford, 1900. [September 16, 1900.] 
