Gravitational settling is widely accepted as being a fundamental physical process acting upon superficial layers of white dwarfs and resulting in an important alteration of their atmospheric composition. Several investigators have been interested by the problem of gravitational settling in white dwarfs (Fontaine and Michaud 1979;Vauclair, Vauclair, and Greenstein 1979; Alcock and Illarianov 1980; Muchmore 1984;Paquette et al. 1986). As pointed out in Paquette et al. 1986, they all reached the same qualitative conclusion: the gravitational settling time scales of metals in cool white dwarfs are small compared to their evolutionary time scales. These stars should therefore have their photospheres depleted of metals if there is no extrinsic source such as accretion for example. This is consistent with the observational fact that most of the cool white dwarfs spectra just show hydrogen and helium lines while the absence of metallic lines indicates a strong depletion of metals. Although the qualitative agreement between theory and observations is satisfactory, only time-dependent calculations can lead to a thorough understanding of the heavy element abundance patterns in cool white dwarfs. In particular, the predicted abundance of an element within the framework of the accretion-diffusion model does depend explicitly on the results of such calculations. We have already presented some preliminary results of numerical simulation of accretion episodes of heavy elements into white dwarfs (Dupuis et al. 1987). As part of an ongoing detailed investigation of these processes, we focus here exclusively on the mechanism of gravitational settling in white dwarfs in order to clear some confusion which has appeared in the literature.The usual notion of an e-folding diffusion time scale strictly applies at the base of the convection zone of a stellar model after having made a number of approximations. If an element can be considered as trace and if ordinary diffusion can be neglected (i.e. the concentration gradient is much smaller than its equilibrium value; see Fontaine and Michaud 1979), then it is a simple matter to show that, at the base of a convection zone, the abundance of a given element is given by,where 8, the diffusion or settling time scale, is given by,where G is the gravitational constant, g is the local gravity, q is the fractional mass, p is the density, and u is the diffusion velocity of the trace element with respect to the background (w < 0). The exponential behavior shown in equation (1) is not valid in regions below the convection zone and the notion of an e-folding diffusion time scale loses its significance there. The time-dependence of the abundance at a given shell can only be obtained from a solution of the continuity equation and, generally, the behavior is very different from an exponential 359 available at https://www.cambridge.org/core/terms. https://doi