230 ON A DIAGRAM OF FREEZING-POINT 
close inspection, but both are clearly consistent with the 1.85 
value of the depression constant. If the leftward bending of the 
KNO, curve in its lower part were actual, as well as the position 
of the curve, the formation of triple molecules might be indicated. 
But being a Loomis curve, it is open to the suspicion of being as 
a whole, too far to the left; and it is plotted with doubtful 
coefficients. 
The KOH curves, Loomis’s on the main diagram and Jones's 
on the inset, are useful only to illustrate the difficulty of making 
concordant observations by different methods. As usual, Loomis’s 
is to the left, and Jones’s to the right. 
Electrolytes such as BaCl,, H,SO,, Na,CO,, have 2 equiva- 
Jents in the single molecule, which may dissociate into 3 or into 
2 ions. If there is no association, they will therefore have 2—3 
or 2—2 curves, according to the mode of ionization. If there is 
complete doubling of molecules, the curves will be 4—6 or 4—4 
curves, provided the doubling does not involve change in mode 
of ionization. Otherwise they might be 4—5, 43 or 4—2 
curves. (The corresponding tangent lines are so far to the left 
of the experimental curves that they are not entered on the 
diagram.) If the molecules are associated in threes, the curves 
will be 6—9 or 6—6 curves, with the above proviso. 
Both Loomis’s and Jones’ curves for BaCl, are shown on the 
diagram, plotted with Whetham’s coefficients (rough extrapol- 
ated values, however, in the case of the former). Ponsot’s curve 
agrees very closely.with Loomis’s. Bearing in mind the right- 
ward and leftward tendencies of Jones’s and Loomis’s curves, 
respectively, we may conclude from the curves of the diagram 
that the actual curve runs down to the right of the 2—3 line, 
bending away from it to the right, and that it would intersect 
the a=1 line at a point between the 2—3 (1.85) and 2—3 (1.87) 
intersections, probably nearer the former than the latter. The 
curve is thus, so far as we can judge, a normal 2—3 (1.85 + .01) 
curve, running, however, very close to the 2—3 Jine. The dia- 
gram, therefore, indicates that BaCl, exists in solution in single 
molecules, dissociating into 3 ions, at least for the most part, and 
that it has a depression constant nearer 1.85 than 1.87. 
