102 Mr. J. H. Alexander on the Tension of Vapour of TVater, 



considered, therefore, as offering an impartial, if not favour- 

 able, term of comparison. 



The mean temperatures are given here, as in our former 

 table, in degrees of Fahrenheit; the individual differences 

 between the thermometer readings are, to save calculation, 

 retained in Centigrade degrees. 



Mean difference 020 



corresponding with mean temperature 228°-2 F. ; mean differ- 

 ence 0°-36 F., which is equivalent to an absolute mean error 

 of0°-18F. 



o 



We have, then, for the mean error at 32° F. 0*42 



167 0-33 



228 0-18 



the average of which, or 0-31 



is the probable amount of error, plus or minuSf with which the 

 various series of M. Regnault are still to be considered as 

 affected. 



Such being the error of the experiments, I shall now show, 

 by the following table, the comparative error of the formula. 

 The quantities in the column of differences are considered as 

 on the same side of the equation with the results from the for- 

 mula; those marked + indicate, therefore, the default, while 

 the sign — indicates the excess of the calculated temperatures. 



