Reviewing the above facts, which represent the general phenomena known, it is apparent that (5) and (6) are independent of the mechanism of production of the charge and relate to the observation of the quantity of charge; again that (3) and (4) eliminate the action of “friction” in the normal sense of the term. It is also obvious that the effect is a surface one and hence is very dependent on the cleanliness and general nature of the surfaces involved. Further, it is evident that much more quantitative work must be done before any theory can be fully elaborated or substantiated. It seems opportune, however, to consider briefly the possibility of stating a theory if only to serve as a stimulus to further quantitative work in this field and the present communication is an attempt in this direction.^*

We proceed then to postulate assumptions:—

Examining these assumptions, 4 is more a fact of experience than an assumption; and 3 is extremely probable in view of the greater mobility of electrons because of their smaller mass. 2 seems fairly probable on the lattice theory of the structure of metals, viz., a positive atom-lattice with an interpenetrating electron-lattice. If we consider the boundary surface of such an arrangement, it seems probable that the surface electron-layer will not remain at the mean lattice-distance but will be drawn towards the inside (with consequent distortion of the surface). The exact surface configuration will depend on the relative positions of the positive ion and the electron-

lattices but it seems fairly probable that there will be some external field.

Assumption 1 also seems fairly probable for “solid” dielectrics. Apparent solids fall into two classes—crystals (the only real solids) and quasi-solids, such as glass which in reality are liquids of great viscosity. In crystals we have the lattice structure which at the surface will undoubtedly give surface-fields within a distance of a few molecular diameters. In the viscous liquids, the assumption is more doubtful but it can be pointed out that the Debye theory of dielectrics,^* which has met with great success in the case of gases and dilute solutions, assumes permanent dipoles in the dielectric (these being oriented by an applied external field) thus giving polarisation in the dielectric. As Debye points out, some such dipoles are really to be expected as the normal state of matter since their absence would imply absolute symmetry in the particles and in the fields on the particles. Adsorption of all gases at the surface of glass seems to point to some surface-field as postulated.

In general, there is some considerable evidence for surface-fields of the type postulated. The work necessary to remove electrons from a metal surface necessitates such a field; contact potentials and the Peltier effect indicate variations of this field from metal to metal. The surface-energy of a crystal-face (varying with the face) must be related to the surface-fields. The alteration in the contact of mercury and glass (from convex to concave meniscus in mercury) when all adsorbed gases are removed, shows the importance of adsorption in this case and as mentioned above, the adsorption phenomena generally would support such a postulate. It is found that surface-films, one molecule thick, on liquids have definite orientations of the molecules. And, finally, on Laplace's theory of surface-tension there exists a force normal to the surface; if this, as seems probable, is electrostatic, then it will orient any natural dipoles in a solid-liquid, so that they are end-on at the surface (such a picture being the “mean” picture, exclusive of thermal agitation).

On the basis of these assumptions, we can picture the phenomena occurring when a flat solid dielectric is brought against a flat metal surface. There are the films of adsorbed gas separating the two; pressure will displace some of these molecules but only those at the “contact spots.” There are thus few places of contact, but if the surfaces are rubbed together, the area of contact is greatly increased, the adsorbed “gas” film being displaced sideways. When the surfaces are within molecular distance (of the order of 10^-7 cm.) the surface-fields become operative (beyond this the field falls off, becoming zero at greater distances since the bodies are as a whole neutral). The exact phenomena of operation of the fields will vary with the particular cases; two fields in the same direction will cooperate; but if opposed, the stronger will determine the direction of electron transfer. If the electron goes to the metal, this acquires a negative surface charge and leaves the dielectric positive (on the surface); the electrons will spread over the metal-surface; the surface of the dielectric will be strained (e.g., if a crystal, the lattice

must be disturbed). The converse occurs if the electron transfers in the opposite sense. In either case the effect is a surface one-the quantity entering into consideration is the surface density and there will be a maximum density determined by (1) the dielectric strength when this is a limiting factor (experiments done by Mr. Macky in the Physics Laboratory here have indicated the importance of this limitation in almost all, if not all, quantitative work hitherto done on the subject), (2) the surface-density of charge necessary to neutralise the surface-field; at least on those parts operative (in the lack of absolute flatness). We might anticipate in this case an increasing density of charge with flatness of the surface, until with absolutely flat surfaces and no dielectric film in between, we should approach a surface-density of the order of that existing, by postulation, owing to natural polarization of the surface.

Obviously since the effect is a surface one, the rest of the specimens do not matter, i.e., the density obtainable is independent of the capacity of the metallic specimen. Obviously also the pressure necessary is only that for contact at the contact spots (where the actual pressure, i.e., thrust per unit area, is probably high); the relative unimportance of pressure is thus easily understood. Again the electron transfer when it does occur probably occurs very quickly and we can see the reason for the absence of any velocity effect in the rubbing.

The so-called frictional series will represent the relative actions of the surface-fields, e.g., if we state a unit F and designate a field as positive when in the direction of the outward normal, then a substance + F will become negative with respect to one with — F; 3F will be negative to 2F, and so on, if we assume equal binding of the electrons in each case. Actually, the resultant effect will be due to the difference of the algebraic sums of the external force and the binding force in each surface, e.g., if an electron is held with f′ and is acted on by F, then (F — f′ is the ejecting force. Thus the electron transfer is determined by (F — f′) — (F′ — f), i.e., (F — F′) — (f′ — f). It should be of interest to work with crystal structures and build up a predicted series.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Let us consider the external surface-fields and their action. In the case of the metal and of a true solid (a crystal), we have the fields due to lattice-structures. These can be evaluated fairly simply, once the surface-deformation is assumed, i.e., the resultant movement due to the surface-discontinuity. Without taking this into account, it is easy to calculate that the surface-field is due mainly to the outermost layer. For example, let us suppose a structure on the cubical system with every particle carrying the same charge e at an equal distance d from its neighbours, and such that planes parallel to the surface are at the same distance d, and alternatively positive and negative. If we suppose the surface-plane positive and consider a point P distant d from this plane, then the total force at P comes out as about 3.5 e/d², whereas that due to the top plane only is 4 e/d². If e is the electronic charge and d = 5 × 10^-8, F = 5 × 10^-10 × 4/25 × 10^-16=c.10⁶ E.S.U., an enormous field = 300 × 10⁶ volts/cm.