Prof. H. L. Callendar on Platinum Thermometry. 219 



As an illustration of the convenience of this method of re- 

 duction a table is appended giving the values of the constants 

 at low temperatures for the specimens tested by Dewar and 

 Fleming. The data assumed in each case are (1) the value 

 of the fundamental coefficient c given in the first column, and 

 (2) the value of the temperature of the O.B.P. on the scale 

 of each particular metal, calculated from the observed re- 

 sistance by formula (1), and given in the third column. The 

 value of the difference-coefficient d° for each metal as deduced 

 from the O.B.P. is found at once by the relation 



^°=(_^-182-5)/5-16. 



The sign of this coefficient indicates the direction of the cur- 

 vature of the temperature-resistance curve, and its magnitude 

 is approximately proportional to the average relative curvature 

 over the experimental range. 



The values of the coefficients a and b, given in the last two 

 columns, are readily calculated from those of c and d by 

 means of the relations already given (p. 199). These co- 

 efficients refer to the equivalent resistance - formula ( 3 ), 

 and are useful for calculating the specific resistance at any 

 temperature. 



In comparing the values of d°, given in this table, with 

 those deduced from observations at higher temperatures, it 

 will be noticed that they are in most cases algebraically 

 greater, the difference amounting to nearly 30 per cent, in 

 many cases between the values deduced from the O.B.P. and 

 the S.B.P. respectively. It is possible that this indicates a 

 general departure from the exact parabola requiring further 

 experiments for its elucidation. It would be unsafe, however, 

 to infer from the results of the present investigation that this 

 is always the case, because, owing to the construction of the 

 coils with silk and ebonite insulation, it was impossible to 

 test the wires directly in sulphur, and they could not be 

 annealed after winding at a higher temperature than 200°. 

 It is well known that annealing produces a marked effect on 

 the form of the curve and on the value of d*. It is also stated 

 in the paper that trouble was experienced from thermoelectric 

 disturbances, owing to the use of thick copper leads 4 mm. 

 in diameter. Such effects cannot be satisfactorily eliminated 

 except by the employment of a special method of compensa- 



* With reference to this point it is interesting to remark that Messrs. 

 Heycock and jNeville with one of their perfectly annealed pyrometers of 

 pure wire, for which c = '00387, e7= T497, found the value pt=— 80 o, 3, 

 t=— 78 0, 2C, for the C0 2 B.P. This would perhaps iudicate that the 

 larger values of d were due to imperfect annealing. 



