the Thermodynamic Scale of Temperature. 137 



deviation of the substance in question from the behaviour of 

 a perfect gas. 



M. Chappuis has recently investigated the thermal properties 

 of several gases; and he has made a careful determination of 

 the increase in pressure under constant volume due to a rise 

 of temperature from the freezing-point to the boiling-point. 

 (Travaux et Memoires du Bureau International, tome vi. 

 p. 108 and p. 78.) We may take his results as enabling 



us to fix the value of — ^— {t l — 1 ) for hydrogen and 



Pi — 'Po 

 nitrogen. In order to obtain t we must calculate the cor- 

 rections according to the formula last written; for this purpose 

 the values of a x and « 2 , given in the preceding Section, may 

 be employed. We thus obtain the following results : — 



Pressure at Uncorrected 



freezing-point estimate of Cor- Corrected 



inM. Ckappuis's temperature of rection. estimate, 



experiment. freezing-point. 



Hydrogen . . 100 cm. of mercury, 273-031 -119 273-153 



Nitrogen.. 99'5 cm. of mercury. 272-131 1*227 273-361 



The two estimates of t are seen to differ by *208, a quantity 

 large enough to be distinctly disappointing when we consider 

 the great care which has been taken in observing the experi- 

 mental data leading to the two estimates respectively. An 

 interesting question arises as to the cause of this difference : 

 but at present we do not possess sufficient knowledge to enable 

 us to locate certainly the source of error. The wisest plan 

 appears to be to admit our ignorance in this respect ; and to 

 admit, consequently, that any one of the four quantities which 

 go to make up our two estimates of t may be affected by an 

 error approaching o, 208 C. owing to some unknown physical 

 cause. 



Thermodynamic Correction for Constant- 1 r olume 

 Gas- Thermometer. 



We have now to consider the second of the two problems 

 mentioned in the beginning of the last Section ; viz., how to 

 rind the value of the interval between the freezing-point and 

 any temperature fixed by a definite physical phenomenon. It 

 is well-known that approximately we may find this interval 

 by the use of the constant-volume gas-thermometer. We 

 have still to discover the corrections needed to convert the 



