the 'liter modynamic Scale of Temperature. 135 



The value of ( -r-) at 50° C. employed above for hydrogen 



is not actually given in this form by M. Amagat. That 

 physicist examined the relation of the pressure to the volume 

 of various gases when the temperature is kept constant; and 

 he exhibited his results by plotting the values of pv against/?. 

 He found that within the limits of temperature of his expe- 

 riments the isothermals for hydrogen could be treated as a set 

 of parallel straight lines. This conclusion cannot be accepted 

 as being absolutely correct, since hydrogen can be liquefied ; 

 and we may rather infer that M. Amagafs law is only an 

 approximation. M. Amagat considers that the value of 



(£)> 



. '00078 when the unit of volume is the volume 

 \dp 



occupied by the gas under standard conditions ; and we shall 



probably run the least risk of serious error by attaching 



this number to the isothermal of 50° C. 



In the case of nitrogen a difficulty of another kind arises. 

 We notice that the cooling effect per 100 inches of mercurv 

 is '57b* at 9P-415 C. and '691 at 9l°*965 C. Having regard 

 to the general character of the results obtained by Joule and 

 Kelvin, it seems most improbable that so large a difference in 

 the cooling effect can be due to so small a change in the tempe- 

 rature of the entering gas. It seems more likely that we have 

 here to deal with a serious experimental error, and that one 

 of the two figures just quoted ought to be rejected. I have 

 rejected the figure belonging to the experiment at i)l°'9()5 C, 

 because we find that the entering gas in that experiment was 

 not composed of pure nitrogen but contained a large admixture 

 of oxygen ; and the figure for pure nitrogen finally given was 

 obtained by a large extrapolation. In the experiment at 

 9l 0, 415 0., on the other hand, the admixture of oxygen was 

 very much smaller, and the resulting figure for pure nitrogen 

 is therefore more trustworthy. 



The large number — three — of constants which have been 

 found necessary to reproduce the experimental values prevents 

 our regarding the resulting formulae as being anything more 

 than empirical. This is not a drawback of any importance 

 as long as we confine ourselves to temperatures lying between 

 the freezing-point and the boiling-point of water ; but it 

 would be hazardous to try to extend the formula to tempe- 

 ratures lying far outside these limits. 



The Absolute Value of the Freezing-Point of Water. 



Suppose that we have a temperature fixed by some definite 

 physical phenomenon ; and suppose that we desire to find its 



