BANCROFT. — TERNARY MIXTURES. 331 



The measurements in the tables are the mean of at least four 

 determinations, and the error is probably not more than b% except in 

 the cases where the quantity of one component is less than 0.20 c.c, 

 when it may easily rise to 10%. The values for n are accurate 

 to within 2% without much question. The values for log C are 

 more untrustworthy, being much affected by a slight variation in ?«, 

 while the term G is liable to even greater fluctuations, and is not 

 given, as being too uncertain. Under the headings " Calc." are the 

 values required by the formula to correspond with the experimental 

 data for the other component. The figures in the column marked 

 logC are Briggsian logarithms. As will be noticed, I have not always 

 taken the mathematical mean of this column as the value of log C in 

 the formula. It seemed better to take the value which best satisfied 

 the experimental data, and to ignore numbers which were obviously 

 faulty. 



TABLE I. 



X c.c. H2O; y c.c. CHCI3; 5 c.c. Alcohol. Temp. 20° 

 Formula x«i y=:C-^; n-^ = 1.90 ; log C]_ - 1.190. 



Water. CHCl, 



Average, 1.190 

 Formula x\f^ - Co; n^ — I-IH; log Cg = 0.742. 



3.00 



1.99 



1.01 



0.92 



0.755 



0.635 



0.55 



0.48 



0.43- 



0.20 



0.127 



