COMPARISON OF PLATINUM AND GAS THERMOMETER*. 



133 



0'090 than 0'082, agreeing in a remarkable manner with the result we have just 

 deduced from REGNAULT'S experiments. 



As the mean pressure of our sulphur point determinations was below 760 millims. 

 in all the series, we thought it of interest to recalculate the results of each set, 

 applying the value 0'088 for dtjdp. Combining this with the known value of d .pt/dt, 

 we have for d.pt/dp the value 0'0773 at 445. 



The values of pt, from the separate experiments with K.8 and K.9, are given in the 

 following table : 



These values of pt, only differ very slightly from those previously found. 



We next proceed to find for each thermometer from the equivalent values of T and 

 pt given by the comparisons near the sulphur point the T, corresponding to the value 

 of pt. deduced above. We formerly used for this purpose a formula containing the 

 term (pt pt.) to the first and second powers, but as there appeared some doubt as 

 to how the result might be affected by stopping short at the second term, in the new 

 calculation we tried several formulae of different types, and included varying numbers 

 of experiments in the neighbourhood of the sulphur point. 



We had already satisfied ourselves that CALLENDAR'S formula closely represents 

 the divergence between the platinum and gas scales over the range covered by our 

 experiments. Utilising this formula and including for K.8 all the experiments between 

 T = 412'65 and T = 455'54, nine in all, we obtain for the T. corresponding to the 

 pt, above given the value 445"27, which is sensibly identical with that previously 

 found. For the two series with K.9, however, we find that while the first series of 

 observations gives a result for T, 445'27, the second series, including the comparisons 

 between T = 405'93 and T = 450'58, gives 445'05, which is appreciably lower than 

 the result given in the text. The discrepancy between the two values furnished by 

 the thermometer K.9 is lessened by excluding some of the comparisons which are at 

 some distance from the sulphur point, but the mean result is hardly sensibly affected. 



We have also made the same kind of calculation of a value for T f from the com- 

 parisons with thermometer K.2, though, in this case, none of the comparisons were 

 made at temperatures very near the sulphur point. We find, employing the same 



