Knowledge of the Specific Volumes of Steam. 21 



less than that at 15°. And it is to be noticed that three of the 

 seven deviations in the experiments cited are positive. From 

 precisely 32 on, however, the calorimeter measurements have 

 all very nearly 4°*5 for the lower temperature and 18° for the 

 upper, and the quantities of heat are therefore expressed in 

 terms of c n . Accordingly the total heats from 32 on are 

 about 0*2 per cent too small and should be increased by that 

 amount. This greatly smooths out the break at that place. 



This point is also verified by experiments 8 to 23. In 8 to 

 13 the calorimeter range was 12°*5 to 24°, hence the unit is 

 <? 18 , and the total heats should be a little large ; such is seen to 

 be the fact. In 13 to 23 the calorimeter range was 6° to 18°, 

 the unit accordingly c ia , and the quantities of heat are too 

 small. We find in fact that all the deviations except one 

 are negative. 



The first seven experiments do not bear this out quite so 

 well. Experiments 1 to 4 are in about c 16 as a unit, experi- 

 ments 4 to 8 in about c 13 , and hence the latter should show a 

 slightly greater negative deviation of perhaps 0*1 per cent. 

 This is not so, but the experiments are few in number and the 

 deviations slight. 



It should be pointed out here that it is not assumed that 

 Regnault's formula is absolutely correct, but it is simply used 

 as giving a smooth set of values from which to estimate the 

 concordance of the experiments. 



The writer considers that the facts here brought forward 

 still further substantiate the essential correctness of the com- 

 parison of Regnault's calorimeter temperatures with the air- 

 thermometer. Accordingly he has corrected Regnault's total 

 heats so that they should all be expressed in the same unit, c 16 , 

 and has given in the fourth column of the table on page 20 

 the deviations of the total heats so corrected from those yielded 

 by Regnault's formula. It will be seen that the formula actu- 

 ally fits the experiments much better than it appeared, the 

 greatest actual deviation being about 0*5 per cent. This means 

 in the latent heats a maximum deviation of about 0*7 per cent. 



Experiments 1 to 57 indicate that the formula gives values 

 slightly too great. At 57 there is another peculiar jump of 

 about 0*75 per cent. The explanation of this is not so evident 

 as in the case of number 32. It lies in the fact that as the 

 temperatures became higher the experiments became more 

 difficult to conduct. Regnauit himself says (Memoir, p. 700) 

 that after number 60 the apparatus required frequent repairs 

 on account of the high temperatures. It therefore appears 

 that the highest set of experiments, from 57 on, are of less 

 weight than the others. If we take the ten experiments com- 

 mencing with 47 which lie between 170° and 176° we find for 

 their average fractional deviation — -0043, the maximum being 



