Water dissociation:
H
2O -> H
+ + OH
-
Kw is equilibrium constant for this reaction, so it should look like
Kw = [H
+ ][OH
- ]/[H
2O ]
It is usually assumed that [H
2O ] doesn't change, so we use simplified formula
Kw = [H
+ ][OH
- ]
Both concentrations can be measured which allows determination of Kw constant, Kw = 10
-14.
But if you decide to not ignore possible changes of water concentration you have to use full reaction quotient (you may use determined value of water ionic product in the numerator):
Kw = [H
+ ][OH
- ]/[H
2O ] = 10
-14/[H
2O ]
[H
2O ] = 1000/18/1L = 55.5M
(where 1000 - mas of water in 1L, 18 - molar mass)
If so
Kw = 10
-14/[H
2O ] = 10
-14/55.5 = 1.8*10
-16
pKw = -log(1.8*10
-16) = 15.7
For most practical applications assumption that water concentration doesn't change is good enough - we rarely use dissociation constants determined with better precision than 2 significant digits, and with precision of 2 SD water concentration doesn't change even for 1M solutions - where thermodynamic effects are so strong that results of equilibrium calculations are already dubious (see
ionic strength and activity coefficients lecture at my site). But in case of very precise potentiometric measurements (and Kw is determined for different temperatures with at least 4 SD accuracy) changes in water concentration should be easily visible in the results.