346 PROCEEDINGS OF THE AMERICAN ACADEMY. 



Formula (x - 0.029 j/)"^ {y - 0.093 .r) = Q ; 714 = 3.00 ; log C4 = 2.135. 



Water. Ethylacetate. 



Calc. Found. Calc Found. log C^. 



0.25 0.25 2.01 2.00 2.134 



0.25- 0.25 2.48 2.50 2.138 



0.286 0.285 3.01 3.00 2.134 



0.29 0.29 5.00 5.00 5.1.35 



Average, 2.135 



In this set of tables, as in the first set, the amount of one non-miscible 

 liquid which will dissolve in the consolute liquid decreases as the 

 quantity of the other non-miscible liquid increases. In this case, 

 however, the non-miscible liquids are saturated solutions and it does 

 not follow that the quantity of one pure liquid decreases as the other 

 increases. There comes a point where the rate of increase of one 

 component in the solution in which it is solute is greater than its rate 

 of decrease in the solution in which it is solvent. If we take the 

 general Equation (17), 



(x — So yY (y — s^x) = C^, 



it is obvious that as x increases y will first decrease, pass through a 

 minimum, and then increase. If the same equation expressed the two 

 equilibria, the point where y was a minimum would be the point where 

 the solution is no longer sensitive to an excess of x. In general, the 

 equilibrium for this second stage is given by a second equation, and 

 all we can say in our present knowledge is that at the intersec- 

 tion of these two curves y should have a minimum value. This 

 does not seem to hold in Table XII., where the amount of ethylace- 

 tate soluble in 1 c.c. methylalcohol in presence of 2.50 c.c. water is more 

 than will dissolve when either three or four cubic centimeters of water 

 are added. I am inclined to attribute this to experimental error, as I 

 do not see how there can be two saturated solutions of the same sub- 

 stance in the same solvent. Such a case would be entirely new, and 

 would involve such consequences that it is not to be assumed on the 

 strength of a variation of two one-hundredths of a cubic centimeter on 

 measurements where the probable error is known to be very large. I 

 propose to repeat these measurements on a larger scale, so as to deter- 

 mine what the facts really are. There are also one or two things in 

 respect to the ether-water-methylalcohol curve which need to be gone 

 into more closely. In Table XIV. I give the values for log C when 

 cleared of the term for z, and the values for log A''when the effect of the 

 exponential factor has been eliminated. Both log (7 and log A' are 



