It has been observed over the past 15 years that experimental frequency-dependent dielectric constants of broad classes of materials including polymeric systems and glasses may be interpreted in terms of the Williams-Watts polarization decay function 4a(t) = exp[-(t/T) ¶, 0 < a < 1.The exponent a and the time constant T depend on the material and fixed external conditions such as temperature and pressure. We derive this form of 4.(t) from the following randomwalk model. Suppose that an electric field has been applied for some time to a medium containing many polar molecules (or polar groups in complex molecules) and the direction of their dipole moments remains frozen as the field is removed. Furthermore, suppose that the medium contains mobile defects that on reaching the site of a frozen dipole relax the medium to the degree that the dipole may reorient itself. If the diffusion of defects toward dipoles is executed as a continuous-time random walk composed of an alternation of steps and pauses and the pausing-time distribution function has a long tail of the form qi(t) a: t-"-, then the relaxation function has the above fractional exponential form.In the theory of dielectric relaxation, one writes the frequency-dependent dielectric constant, E(4,)), asEs -Exo°w here EX is the high-frequency limit of the dielectric constant and Es, the static dielectric constant. The function 4(t) describes the decay of polarization of a dielectric sample with time after sudden removal of a steady polarizing electric field. One generally writesEs -Eoc where En(w) and E'"(w) are, respectively, the real and imagi- [4] The En(w) data of Ishida and Yamafugi (3) on polyvinylacetate at 62.5°C was identified with 4a(t) with a = 0.56 over five frequency decades. The Williams and Watts group, Moynihan and his collaborators (4, 5), Ngai and White (6), and other investigators have found the Williams-Watts function 4 to represent a "universal" model for a wide class of materials, especially polymeric substances and glasses. The remarkable empirical success of the Williams-Watts relaxation function motivates one to seek a physical model to give some intuitive understanding of it. The purpose of this paper is to discuss such a model. Suppose that we have applied an electric field for some time to a medium containing many polar molecules (or polar groups in complex molecules such as polymers) and that the medium has relaxed around the polar groups so that the dipole moments freeze in direction after the field has been removed. Furthermore, suppose that the medium contains defects that, through thermal excitation, become mobile, some reaching frozen dipoles and on doing so relax the medium in the neighborhood of the dipoles so that they may reorient themselves as required in an approach to equilibrium. In polymeric materials, the defects may be local conformational abnormalities induced by interaction of a polymer chain with itself or with neighbors, thus introducing local strains into a system. In glassy systems, often with small a va...