Knowledge of the Specific Volumes of Steam. 19 



Number of Extreme 



Observer. Temperature. Experiments. values of L. Mean L. 



Andrews 100 8 530-8-543-4 535'9 



Favre and Silbermann, 99-81 3 532-59-541-77 535*77 



Berthelot 100 3 535-2-537*2 536'2 



Schall 100 532-0 



Fuchs 100 536-8 



Rarnsay and Marshall, 100 537*0 



Regnault _ 99'88 44 533 0-538-0 536-3 



The greatest deviation of these mean values from Regnault's is 

 about eight-tenths of one per cent. To get Regnault's latent 

 heat from his total heat I have used the ' heat of the liquid ' 

 from the formulae already developed. 



Consider now Regnault's experiments on total heats. These 

 are divided into four classes, those at or near 100°, those from 

 119° to 195°, those from 63° to- 88°, and last the set from -2° 

 to +16°. He has given as a formula to unite these 

 H = 606-5 + 0*305 t. 



Take the series at 100°, forty-four in number. Neglecting 

 the first six, which Regnault does not regard as so accurate as 

 the others, the mean is 636*67, and the greatest deviation from 

 this is less than three-tenths of one per cent. If we admit the 

 first six the greatest deviation is less than six-tenths of one per 

 cent. When we consider the fact that Regnault purposely 

 varied all his experiments in many ways so far as possible, in 

 order to eliminate constant errors, this close agreement in so 

 many experiments shows that the mean value given above is 

 very exact. 



The unit in which this is expressed is determined by the 

 calorimeter range, the average of which was from about 8°*3 

 to 20°*9, from which range the individual experiments did not 

 differ essentially. Accordingly the unit is the specific heat at 

 14°*6 according to Regnault's thermometric scale, or practically 

 Regnault's calorie. This we have judged to be the specific 

 heat at 15° according to Rowland's scale, and have investigated 

 the possible error in that judgment. There will be accordingly 

 the same possible error in the total heats. 



Consider next the experiments above 100°, seventy-three in 

 number. A comparison of the numbers given in Regnault's 

 tables with those given by his formula follows : From 119° to 

 150° the deviations of the experiments from the formula vary 

 from —0 - 36 per cent to +0*25 per cent, mostly negative, but 

 are generally very slight. At 153°, or experiment number 32, 

 there is a sudden jump to —0*5 percent, and a deviation exists 

 varying from —0*2 per cent to —0*65 per cent, averaging 

 — 0*5 per cent, and gradually increasing up to 175°*5. Here 

 there is another sudden change, and the deviations vary up to 



