18 H. C. COOKE 
probably reached in a few hours; but owing to delays the solu- 
tions remained for 65 days before they were analyzed. They were 
shaken at intervals during this time. In the analyses, the ferrous 
salt was first determined by standard permanganate; great care 
being taken to stop the titration as the first faint tinge of pink 
appeared. The silver was determined in the same sample by 
titration with ammonium thiocyanate. The method was found 
very accurate, duplicates checking to one-tenth of 1 per cent. 
The ferric salt was determined by difference. In calculating the 
constant from the results obtained, it was assumed that the salts 
were all completely ionized in the dilute solutions used. This 
assumption is scarcely permissible; but the writer was unable to 
obtain any figures on the ionization of ferric sulphate in solutions 
of different strengths and for the other salts the difference between 
complete ionization and their ionization in these solutions is small. 
Concentrations in the following table are all given in formula weights 
per liter. 
Fe2(SO;)3 FeSO, Ag2SO, K, 
0.00370 0.03972 0.000439 DD) it 
.00475 .03370 . 000703 200.4 
.00556 .02876 . 000952 203.2 
.00026 .O2451 .001298 196.7 
.00077 .O2TO1 .001708 188.8 
.00695 .01849 .002204 163.5 
.00718 .O16T4 .002738 TO2 RS 
.00737 . 01406 . 003226 162.5 
.00748 .O1231 .003715 163.7 
- 00749 .O1094 . 004200 163.0 
From the results of the last five analyses the constant would be 
163. The higher results of the first five analyses are ascribed to 
increasing hydrolysis of the ferric salt as its dilution increases. This 
would make the amount of the ferric zon in solution less than the 
total ferric value. It is the latter which is measured, while only 
the ionized portion enters into the equilibrium. Therefore, since 
tI} 
F ' 
Cn? XOa! x Ca! the numerator is too large, the value 
e g 
of the fraction must likewise be too large. If this is the true 
* Volhard, Liebig’s Ann. d. Chem., CXC, 1. 
in the fraction 
