The mechanism of flotation of oxide and silicate minerals was established long ago as being due to the electrostatic attraction between the charged surface and the charge of the collector. What is less well established is the mechanism of the charging of the mineral surface. Most researchers have postulated that the adsorption of H+ is the cause. However, the adsorption model and its derivatives (such as the amphoteric and multisite-adsorption models), do not successfully describe the data for the zeta potential as a function of pH. In particular, these adsorption models have the following features that are not observed: (i) they are symmetrical about the point of zero charge, whereas the data is asymmetrical, (ii) they have an inflection at the pK(a) values, leading to an asymptotic approach to the point of zero charge, whereas such an inflection is not observed, and (iii) they only fit the data in its extremes, that is, when the pK(a) values differ by orders of magnitude, or are very close to one another. An alternative mechanism of charging is proposed here that is based on the dissolution of ions from and deposition of ions onto the mineral surface. It is shown that this model (a) fits the data, (b) is consistent with the thermodynamic model for reversible interfaces, (c) is consistent with the thermodynamics of the overall reaction and its solubility, (d) indicates reasons for the ageing of surfaces and (e) displays the observed features of zeta potential measurements as a function of pH, that is, it is pseudo-Nernstian and asymmetrical in nature. Application of the proposed theory is discussed for the flotation of quartz and corundum. (C) 2016 Elsevier Ltd. All rights reserved.