Prof. H. L. Callendar on Platinum Thermometry. 215 



It is generally sufficient to put D = D' on the right-hand side 

 of this formula, so that if either is known the difference 

 between them may be determined with considerable accuracy. 



It will be observed that the table of reduction given by 

 Dickson agrees very closely w r ith either of the difference- 

 formulae. But, on the whole, most closely with (D). If 

 Dickson had calculated his formula from the same data it 

 would have given results identical with (D'). By giving 

 equal weight, however, to all the observations, without regard 

 to steadiness of temperature or probable accuracy, he is com- 

 pelled, as in the previous instance, to admit an error of o, 35 

 in the fundamental interval itself, which is quite impossible. 

 Except at these points the probable error of his reduction is 

 not of vital importance ; on the contrary, the general agree- 

 ment with (D) is so close that it is difficult to see on what 

 grounds he can regard the latter as being either incorrect or 

 inadequate. 



For practical purposes a table of this kind is not convenient 

 owing to the continual necessity for interpolation. A graphic 

 chart in w T hich t is plotted directly against pt is objectionable, 

 because it does not admit of sufficient accuracy unless it is 

 plotted on an unwieldy scale. The difference- curve avoids 

 this difficulty, and is much to be preferred for laboratory work. 

 But for occasional reduction it is so easy to calculate the 

 difference directly from the formula that it is not worth while 

 to take the trouble to plot a curve. 



Reduction of Olszewski's Observations. — The observations 

 of Olszewski on the critical pressure and temperature and 

 boiling-point of hydrogen, described in the Phil. Mag, for 

 July 1895, were made with a platinum thermometer of '001 

 inch w T ire wound on a mica frame in the usual manner. He 

 graduated this thermometer by direct comparison with a 

 constant-volume hydrogen thermometer at the lowest tem- 

 peratures which he could obtain by means of liquid oxygen 

 boiling under diminished pressure. The lower temperatures, 

 observed with the thermometer immersed in temporarily 

 liquefied hydrogen, were deduced from the observed resist- 

 ances by rectilinear extrapolation, assuming that the resistance 

 of the platinum thermometer continued to decrease, as the 

 temperature fell, at the same rate as over the lowest tempe- 

 rature interval, —182*5 to —208*5, included in the range of 

 the comparison with the hydrogen thermometer. It is pos- 

 sible that, at these low temperatures, the resistance of platinum 

 does not continue to follow the usual formula, but it may be 

 interesting to give a reduction of his observations by the 

 difference method for the sake of uniformity of expression. 



Q2 



