Electroencephalography has added another important means of investigation of cerebral activity and functioning to the many others previously at our disposal. In applying it, as with the other investigative methods, it is important that we should have clearly in mind exactly what facts it is able to demonstrate and what interpretations we are justified in making of the observed facts. This process of interpretation is difficult partly because it is a two-stage one; the electrical potentials observed are those on the surface of the scalp and these must first be referred back to potential sources (i.e. electrical charges) in the brain itself. These latter must then be interpreted in terms of activities of nerve cells and nerve fibres. The first of these two stages belongs entirely to the sphere of mathematical physics, and the present paper is an attempt to present a working hypothesis which not only suggests a solution of this stage of the problem, but suggests a probable solution of the second stage. The second stage properly belongs to the sphere of physiology, but progress here is rendered difficult by lack of decisive knowledge as to the nature of the elementary unit of central nervous activity.