Abstract. -Theoretical treatment of the mechanisms associated with thermally activated grain boundary migration requires a knowledge of the atomistic and electronic structure of the grain boundary. Although in recent years a whole range of new data has been acquired concerning the atomistic structure of grain boundaries (see for example the relevant papers presented at this colloquium) they can do no more than qualitatively extend the existing, in some cases really old, heuristic concepts of grain boundary migration mechanisms. The effect of the electronic structure on grain boundary migration has not been considered at all till now. The idea that this aspect also is of importance is rather new.In view of this, and considering that there are already a number of recent review articles on the subject of grain boundary migration [1][2][3][4], the present paper is restricted essentially to the discussion of several unanswered questions which are of general importance for a deeper theoretical understanding of grain boundary migration in very pure metals. The discussion is also restricted to large angle grain boundaries. Orientation effects will not be included because they are discussed in another of the papers to be presented.1. Current theories. -The theories are divided into single process and group process according to the assumed method of materia] transport which effects the displacement of the grain boundary [1]. In single process theories it is assumed that the elementary process depends on the uncorrelated release or deposition of individual atoms from the lattice and on to the lattice of the neighbouring grain. The rate of grain boundary migration has been calculated many times on this basis of this theory [1, 5]. The individual treatments differ only in the way in which the overall process is split into individual stages. In the group process theories it is assumed that a group of atoms change from one lattice to the other in a thermally activated event.Here again various explicit formulations exist for the rate of grain boundary migration [1].In the past, interest has centred primarily on single process theories because it was considered that they would yield better agreement with experimental data for grain boundary migration in very pure metals than group process theories [1,2, 6]. In spite of the success of the single process theories, several fundamental questions have still not been answered completely or have been provoked again by more recent experimental observations. Single process theories yield an expression for the rate of migration of a boundary of the form in which b is the atomic diameter, v the effective attack frequency, p the driving force (energy per atom), AS and AH the activation entropy and enthalpy and is ...