The measured solution conductivities
exp are shown in Table 2,
along with the estimated conductivities
calc calculated
from Eqn. 2. Within the table,
the results are divided among the three molar ratios.
Also shown in Table 2 are the coefficients of
variation
:
![]() |
(8) |
Because the coefficients Gi were not optimized for these three ionic species, all the estimated values lie below the measured values. Optimizing the G may not be warranted because the empirical relation in Eqn. 3 is a coarse approximation. Moreover, the present error is already less than 8 % over the entire concentration range.
| [K |
[Na |
exp (S/m) |
calc
(S/m) |
![]() |
| 0.125 | 0.03125 | 3.707 | 3.591 | -0.031 |
| 0.250 | 0.06250 | 7.133 | 6.796 | -0.047 |
| 0.500 | 0.12500 | 13.56 | 12.64 | -0.068 |
| 1.000 | 0.25000 | 24.78 | 22.99 | -0.072 |
| 0.125 | 0.0625 | 4.346 | 4.201 | -0.033 |
| 0.250 | 0.1250 | 8.330 | 7.913 | -0.050 |
| 0.500 | 0.2500 | 15.60 | 14.63 | -0.062 |
| 1.000 | 0.5000 | 27.61 | 26.45 | -0.042 |
| 0.125 | 0.125 | 5.642 | 5.387 | -0.045 |
| 0.250 | 0.250 | 10.70 | 10.07 | -0.059 |
| 0.500 | 0.500 | 19.57 | 18.45 | -0.057 |
| 1.000 | 1.000 | 33.50 | 33.02 | -0.014 |
The performance of Eqn. 2 is relatively uniform over the range of ionic strengths investigated. The data from Table 2 are plotted in Fig. 2(a) (filled symbols) as a function of the solution ionic strength IM. The predictions from Eqn. 2 are shown as solid curves, one for each of the potassium to sodium ratios. The three curves are nearly collinear, as are the measured values.
For comparison purposes, also shown in Fig. 2(a) are estimates that neglect the concentration dependence of the equivalent conductivity:
| (9) |
![]() |
![]() |
This idealized approximation is mathematically equivalent to setting all the Gi coefficients to zero in Eqn. 3. These approximations are shown as dashed curves in Fig. 2(a), one curve for each of the three potassium to sodium ratios. At an ionic strength of 1.0 mol/L, the idealized approximation differs by 36 % from the experimental data, and differs by 55 % at 2.0 mol/L.
The graph of the coefficient of variation
plotted in
Fig. 2(b) shows that the
relative error is fairly constant over the entire range of ionic
strengths. This suggests that small extrapolations of either the
ratios or concentrations beyond the
parameter space investigated here should not introduce extensive
uncertainty.