BANCROFT. — TERNARY MIXTURES. 363 



The first thing that strikes one about this table is the way in which 

 so many of tlie h + 1 values approximate to 1.40. Why this should 

 be so is entirely unknown. In tiie log K values we notice that, for the 

 same acid, increasing the carbon atoms in the alcohol radical diminishes 

 the constant. There is only one exception to this, butylformiate, and 

 here the possible error is very large. It looks also as if the constants 

 might be additive, being made up of one factor for the alcohol and 

 another for the acid radical ; but the experimental data is too insuffi- 

 cient to Justify this hypothesis. It is very much to be hoped that 

 some one will make a careful series of experiments to settle this 

 point. 



Formula II. was deduced for the case when the reacting weights of 

 the substances in equilibrium are not functions of the concentration. 

 The measurements of Pfeiffer and myself show that, with the possible 

 exception of the chloi'oform-water-acetone series, this condition has been 

 satisfied in all the cases studied, though the experiments extended over 

 a wide range of concentrations. This is in flat contradiction with 

 the determinations of the reacting weights by the boiling-point and 

 freezing-point methods. These methods give accurate results only for 

 very dilute solutions, and even then only for certain solutes in certain 

 solvents. To explain the variations, we are forced to assume "double 

 molecules " in some cases, polymerization with increasing concentration 

 in practically all cases, and " variations from the gas laws." I have 

 brought together a large series of measurements in which there is no 

 sign of any of these things. I see only two jjossible hypotheses to 

 account for this discrepancy : first, to enunciate a new and most in- 

 teresting law, to wit, presence of a third substance prevents " polymeri- 

 zation " and " variations from the gas laws " ; second, the formula for 

 the change of vapor pressure with the concentration is incorrect. Tlie 

 first hypothesis seems to me out of the question, and there remains 

 only the second. It is a bold thing to question so univei'sally accepted 

 a formula, but I feel convinced that it is not right, and that equal react- 

 ino- weights of different substances do not produce the same change of 

 vapor pressure. I think that the mistake in the past lay in assuming 

 that the work done in compressing a dissolved substance from the 

 volume Fi to the volume V^, by means of a semipermeable piston is 

 equal to f p d v between those limits, irrespective of the nature of 

 solute and solvent. I have already collected some experimental evi- 

 dence in favor of this view, and I hope before long to be able to 

 establish my point. 



The facts brought out in this paper throw light on a research by 



