Knowledge of the Specific Volumes of Steam. 23 



It is immediately seen that Regnault's formula gives results 

 uniformly too great. The reason is because the formula is 

 made to fit his experiments near 0°, which we shall see to be 

 at fault. The average deviation of the formula from the 

 experiments is 0'3 per cent. 



The experiments near 0° are undoubtedly incorrect. They 

 were performed by a method different from that which was 

 used in the remainder, and Regnault himself does not con- 

 sider them so good as the others, indicating a number of causes 

 of uncertainty (Memoir, pp. 716-719). Another source of 

 error lies in the fact that Regnault supposed, in common with 

 the physicists of his time, that the heat necessary to form 

 vapor depended only on the final state of the vapor and not at 

 all on the process by which it was formed. This error did not 

 enter into the preceding sets of experiments, for in them the 

 process was always one of sensibly constant pressure. In the 

 present case, in order to produce the evaporation the pressure 

 on the liquid was reduced, and the operation was accordingly 

 not one of constant pressure. 



In place of these unreliable experiments we have the recent 

 accurate work of Dieterici. He found for the latent heat at 

 0° the value 596*8, the unit being <? _ 100 . This is the result of 

 a very careful investigation. He made first a series of eight 

 experiments, L varying from 595*74 to 597'29. He then made 

 a better set, three in number, the extremes being 595*98 and 

 596*63. He still further improved his methods and in two 

 experiments obtained 596*67 and 597*07. The probable error 

 is ± 0*24 or dt 0*04 per cent. In terms of c lb as a unit the 

 latent heat at 0° thus becomes 598*9. Svensson's determination 

 is somewhat larger than Dieterici's, it being 599*92 in terms of 

 ^o-ioo as a um 't j lfc i s probably not so correct as Dieterici's, as 

 shown by the remarkable concordance in the observations of 

 the latter. Both have the slight possible error due to the 

 uncertainty of the constant of the ice-calorimeter, which we 

 have seen to be about one-quarter of one per cent. 



Griffiths' determinations were of the latent heat between 25° 

 and 50°, the determinations at 30° and 40°*15 being made with 

 particular care. They are as follows, the unit being c lb : 



t 



25 



30 



40 



40*15 



49*8 



L 



581*9 



578*7 



572*4 



572*6 



566*5 



These give total heats lying considerably below Regnault's 

 linear formula, in this respect agreeing with the experiments 

 of Dieterici, Svensson, and those of Regnault himself from 60° 

 to 90°. This strengthens the reliability of the formula we 

 have formed for the region above 100°, it also lying below the 

 linear formula. As any expression for the total heat must 



