218 Speyers — Molecular Weights of some 



table was made. However no other way was found of getting 

 at the true error. 



This method of getting the molecular weights of solutes is 

 only useful for solutions which do not change chemically when 

 a current of air is passed through them for twenty-four hours. 

 Moreover, the solutions must be concentrated, because all the 

 errors are piled up on the molecular weight to an extravagant 

 degree when the solutions are dilute, that is, when p — p' is 

 small. The weights of w, W, and M are to be considered cor- 

 rect as given ; then the effect of an error of one unit in the 

 determination of p and of p' gives an error in the determina- 

 tion of m proportional to 



d jt^ d jL^ j 



p—p -f p—p=z - -dp 

 p—p 



dp dp' 



from equation (3), and 



<%£- dl^ r p-p' 



P + P — in i rx a ty 



from equation (4), which whenp—p' is small, becomes 



/ — tv- dp =■ -. 7T dp approximated. 



P{P—P) {P—P) ri J 



A few measurements with dilute solutions are given but 

 they are not to be depended upon, although the results seem 

 good in some cases. For such solutions, the boiling point 

 method is by far preferable. 



The observations and molecular weights are given in the 

 following tables. In the first column is the solute, its ordinary 

 molecular weight, its origin and purity ; in the second, the 

 quantity of solvent in grams ; in the third, the quantity of 

 solute ; in the fourth, the corrected losses of tube 1 - p'f in 

 the fifth, the corrected losses of tube 2 —p; in the sixth, the 

 differences =p—p'; in the seventh, the molecular weight 

 according to equation (3) ; in the eighth, the molecular weight 

 according to equation (4) ; in the ninth, the temperature. 



Let us consider each solution separately. 



Urea in water. 50 per cent to 60 per cent urea referred, as 

 always, to 100 parts of solution. The mean molecular weight 

 of urea according to formula (1) is 60*6 ± 1*6 ; according to 

 formula (2) it is considerably greater, 72*6 =b 1*8. This assumes 

 that we accept the modified van't Hoff theory. If we deny 

 the validity of the modification, then formula (2) cannot be 

 used at all, for the necessary data as to compressibility and 

 heat of dilution are not known. Formula (1) much to be pre- 

 ferred. 



