

1904] Mills — Molecular Attraction. 93 



now be questioned. 1 But it is a purely empirical formula 

 fitted to the observations. It represents a curve and primar- 

 ily is but a more refined method of drawing- a curve through 

 the observations. For all points along 1 the line of this curve 

 one is guided by the observations on either side until the ends 

 of the curve are approached. But as the ends of the curve 

 are approached one has to be guided more and more by the 

 trend of the curve already established. That is, along inter- 

 ior portions of the curve individual errors in the measure- 

 ment of the pressure are smoothed out and Biot's formula is 

 far more accurate than the individual measurements, but at 

 the ends of the curve this smoothing is necessarily far more 

 imperfect and Biot's formula cannot give greatly more accu- 

 rate results than the individual observations. (The fact that 

 the constants for the formula are mathematically calculated 

 makes this observation of course none the less true.) 



Individual observations of vapor pressure could not be used 



8P 

 to obtain a correct idea of the -^7=- because an error in these 



observations is multiplied proportionately anywhere from ten to 



8P 



seventy times in the-^p^. The method adopted will keep this 



enormous multiplication of error from being greatly apparent, 

 except as either end of the Biot formula curve is approached. 

 But throughout the entire range of the observations it is 



ap 



doubtless this multiplication of error in the -«-=- that is most 



often responsible for the variation in the constant. 



Many illustrations taken directly from the measurements 

 enforcing the above remarks might be given, but attention 

 will only be called to the case of water. Ramsay and 



1 See Ramsay and Young. Phil. Trans. 1887A, p. 82. 



