Electrolyte-induced coagulation of colloidal dispersions of silica has remained a puzzle for many decades, and it is widely considered anomalous from the viewpoint of traditional Gouy Chapman theory of diffuse double layer and zeta-potential at ideal surfaces and of their electrostatic interaction (Derjaguin-Landau-Verwey-Overbeek, DLVO theory). It is suggested that this anomaly is caused by the fact that silica particles are covered with swellable gel layers. Theoretical stability ratios are calculated combining the attractive van der Waals and repulsive electrosteric interactions between core shell (soft) model spheres with homogeneously distributed fixed charges in the shells and matched with the experimental ones measured for nonporous silica microspheres of different diameters (50, 150, and 320 nm) in an univalent electrolyte (KCl) of increasing concentration and pH (2.6, 4, 6, and 8). The variation in the shell thickness with the KCl concentration (mimicking the charged gel C layer swelling) as the only adjustable parameter, deduced in such a way from data at pH 6 and 8, not only can explain the parallel experimental electrophoretic mobilities but also conforms itself to a scaling law derived from the thermodynamic theory of polyelectrolyte hydrogels. A resulting inapplicability of the DLVO theory and the zeta-potential concept for a quantitative predicting the coagulation kinetics of gel layer-covered colloids is discussed.