In a recent paper on a model A-B alloy containing a high concentration
of vacancies, Qin and Murch (1995, Phil. Mag. A, 71, 323) obtained 't
hermodynamic factors' for diffusion of the two components. These diffe
red from one another, whereas the usual treatment of chemical diffusio
n leads one to expect them to be equal. In this letter the reasons for
the difference are analysed from two points of view, one purely therm
odynamic, and the other statistical (in the sense that it recognizes t
he presence of vacancies in the system). Within the statistical framew
ork (used also by Qin and Murch), a relation between the thermodynamic
factors for A and B is derived. In the special case of the symmetrica
l model used by Qin and Murch, this relation shows the reciprocity vis
ible in their numerical results.