422 Mr. H. M. Tory on the 



Phil. Mag. Feb. 1899, p. 200) gave results agreeing with the 

 standard wire almost within. the. limits of error of observation, 

 but did not apply so well in the case of the most impure 

 specimens. It was observed, however, that the differences 

 of the platinum temperatures for all the wires could be very 

 closely represented between the limits of observation by 

 linear formulse. If the observed values of pt for each wire 

 were first reduced to the standard pt in each case by the 

 linear relation, and then corrected by the same difference- 

 formula as the standard wire, very consistent results were 

 obtained from all the wires. This method would not be 

 applicable below 400° C, but it might prove a suitable t}q>e 

 of formula for purposes of extrapolation, in case it was 

 necessary to employ impure wires. 



It is probable that the platinum wires used in this investiga- 

 tion were free from contamination with baser metals outside 

 the platinum group. Under proper conditions they all showed 

 great constancy of zero, and the variation of the difference- 

 coefficients (from 1*50 to 1'67) was comparatively small. 

 By contamination with baser or more volatile metals, it 

 is possible to get much larger variations in the values of pt 

 and<^; but such pyrometers will not show constant results 

 until the volatile constituents have been burnt out. Special 

 attention was given to the effects of annealing which may 

 produce changes of zero of the order of one degree after the 

 first exposure to a temperature of 1000° C. With this excep- 

 tion, no changes of zero of any importance were observed, as 

 the temperatures employed never exceeded 1000° C. 



2. Methods and Apparatus. — The method of comparison 

 adopted was that described by Callendar (Phil. Trans. A, 1887, 

 p. 161), in which the wires to be compared are wound side by 

 side in the same tube, so as to be always at the same mean 

 temperature. This method appears to be more accurate than 

 the method of indirect comparison of pyrometers in separate 

 tubes. Observations were taken at steady temperatures only, 

 in a well-stirred bath of melted tin. The apparatus was 

 arranged so that the readings of the two pyrometers were 

 almost simultaneous. ' Each wire was directly compared with 

 the pure standard wire ; and in addition a specimen of the 

 standard was drawn down to half section, and compared with the 

 original wire. The agreement was found to be exact through- 

 out the scale, as was to be expected with so pure a wire. 

 The wires were wound on mica strips with platinum leads 

 insulated in the usual manner. Each double pyrometer, 

 when completed, had six leads, two each for the pyrometer 

 coils, and two compensating leads. The resistance-box used 



