The unknown protonation constants were calculated in two ways. Firstly, using the values of parameters E₀, S, j₁, j₂ determined from the corresponding calibrations. The second approach is based on a simple idea that not only calibrations, but also titrations carry significant information about unknown calibration parameters and that the precision of the potential measurement is the same in both calibration and titration. Thus, all the titrations were processed together and each couple of titrations (i.e. calibration and the corresponding titration) was assigned a joint set of unknown parameters E₀, S, j₁, j₂. Then, the variance uniformly spread over all experimental points (with respect to their statistical weights) and the residuals became smaller. Before applying this approach the chemical model must be known!

Analytical concentrations (mass amounts) of proton and ligand were determined either by weighting of solid compound with "known" stoichiometric composition (ethylenediaminetetraacetic acid and monohydrate of N-glycylaminomethylphosphonic acid), by chelatometric titration with lead nitrate (in the case of ethylenediaminetetraacetic acid only) on the assumption that the solid compound was stoichiometrically protonated, or by nonlinear least squares method (both independently and on the assumption described above applying "natural" linear constraint for unknown mass amounts of proton and ligand, i.e. boundary condition for unknown mass amount of proton).

The computed standard deviations seem to be inadequately small. This fact mainly points to a high precision and good reproducibility of the measurements, the use of a suitable calibration function with many degrees of freedom and also to a large number of acquired experimental points. However, uncertainties of some of the analytical concentrations (mass amounts), volumes, and known stability constants as well as a presumable presence of impurities in the chemicals and stock solutions used or a variability of the ionic strength were not taken into account. That is why the real accuracy is substantially lower (and the real standard deviations are substantially larger). According to the conclusions of several inter-laboratory tests the real standard deviations of the determined protonation and stability constants are at least 0.05-0.1 units in logarithmic scale.