72 Prof. H. L. Callendar on the T her modi; namical 



gas at a slightly lower initial pressure. p = S7 cms. The 

 observed values for the difference in these cases have been 

 increased in the ratio 100/87, as required by theory, and the 

 crosses are distinguished by inclosing them in circles. It 

 will be observed that the corrected observations agree in the 

 most remarkable manner with the Joule-Thomson curve, 

 although their agreement with ChappmV formula is slightly 

 impaired. This would appear to be a striking confirmation 

 of the validity of the proposed formula in the case of C0 2 . 

 But when we compare the actual values of c and b calculated 

 as above from the observations on the cooling- effect alone, 

 with those calculated directly from the slope of the isothermals, 

 we find certain discrepancies which, although they are often 

 within the limits of experimental error, require examination 

 as possible indications of some defect in the theory. For 

 instance, in the case of C0 2 , which agrees so well with 

 Chappuis' thermometric comparisons, the value of c according 

 to Table II. would become equal to that of b at about 120° C, 

 and the gas above this temperature should behave like 

 hydrogen, with an upward slope of the isothermal. The 

 observations of Amagat show, on the other hand, that C0 2 

 is still notably imperfect at a temperature of 261° C. This 

 might conceivably be due in part to some effect of surface- 

 condensation, which would be relatively important in the fine 

 tubes employed by Amagat : but it is mainly attributable to 

 the large value of b deduced from the observations on the 

 cooling-effect. It is evident that the value of b cannot be 

 deduced so accurately as that of c from these observations, 

 since the expression for SQ is 3c— b. Moreover, no account 

 has been taken of the variation of S with temperature, which 

 according to Regnault is considerable in the case of CO.->. 

 Both these considerations would be of relatively small im- 

 portance as affecting the thermometric comparisons between 

 0° and 100°. since b does not enter into the expression for dt, 

 and we employ the mean value of S at 50°; but they would 

 materially affect the extrapolation of the value of c — b. It 

 would be quite possible to readjust the values of c and b in 

 such a manner as to agree better with Amagat at higher 

 temperatures, while not seriously impairing the agreement 

 with Chappuis at 50\ The values of c and b for hydrogen 

 appear to be more nearly of the right order of magnitude, 

 giving c — 6 = 8*2 c. c. as against Amagat's value 8*8 c. c. 

 On the other hand, the value of b for air is practically zero 

 according to the observations of Joule and Thomson, i. e. 

 air would always remain imperfect. Observation shows, 



