Mr. J. H. Alexander on the Temion of Vapour of Water, lOl 



parison other than has been instituted. But the observations 

 at32'' F., a temperature especially disengaged from instru- 

 mental errors, are, as has been already said, very numerous, 

 and allow of being compared among themselves. Of the forty- 

 seven observations whose arithmetical mean pressure is given 

 in the table, the 



ill. o 



maximum was 0-18485 ; corresponding to a temp, by formula of 30-72 F. 

 and mmjjwMOT was 0-17717 29-77... 



and the difference 0-00768 



corresponding difference of temp. 0-95 



This difference shows a mean error in temperature, unac- 

 counted for, of 0°*425 F. ; and a limiting error in pressure 

 rather more than half the difference between the formula and 

 the mean of all the observations. 



In the various series of M. Regnault, the temperature is 

 given sometimes by one thermometer only, and sometimes by 

 two, and even four. Of these latter classes, I have taken out 

 of each series the observation where the difference of reading 

 of the several thermometers is the greatest, to serve for another 

 comparison, as follows : — 



This difference is that of the extremes ; and as the mean error 

 of any number of observations is as likely to he plus as mivjis^ 

 it is equivalent in this example to an absolute error of 0°'33 F. 

 This error manifests itself in a series where the thermometric 

 variations are the greatest. I shall now present another where 

 these same differences, although not perhaps the lowest of all, 

 are yet very much less than in the last. At least, this series 

 (which in fact forms part of the comparative table in the pre- 

 ceding memoir) was selected without any reference to the 

 present investigation, and with a view to the introduction of 

 the greatest number of accordant observations, and may be 



