. In many objects, the turbulent particle diffusion coefficient is at most 10 times larger than the atomic diffusion coefficient.
ContextOver the last few years the quality of observational material has improved to the point that much finer tests of stellar evolution models have become possible. One example is the dip in Li abundance at T eff = 6700 Κ reviewed by Rebolo elsewhere in this volume. Demonstrating the presence of the Li gap has been made possible by the high signal to noise data available on relatively faint objects.We develop tests of the physical processes using the observed abundance anomalies. In addition to "standard" stellar physics, we shall consider particle transport by atomic diffusion, meridional circulation, mass loss and turbulence. Atomic diffusion is a basic physical process that must be included unless there are more efficient transport processes that compete with it and wipe out its effects. Meridional circulation is calculated from first principles without arbitrary parameters (Tassoul and Tassoul 1982); it defines a model for particle transport in rotating stars whose effects can be tested. Mass loss is known to be present in the Sun at a rate of 10" 14 M 0 yr" 1 and it is appropriate to determine what would be its effect in other stars at that level. While mass loss is a natural phenomenon that may well play a role, its introduction currently requires at least one arbitrary parameter. As to turbulence, it is discussed elsewhere in this volume and we shall mention it only in so far as it is possible to put interesting upper limits on it.