A knowledge of these things would amount to at least the beginning of a complete dynamical theory of electricity, in which we should regard electrical action, not, as in this treatise, as a phenomenon due to an unknown cause, subject only to the general laws of dynamics, but as the result of known motions of known portions of matter, in which not only the total effects and final results, but the whole intermediate mechanism and details of the motion, are taken as the object of study. Maxwell did not always express himself so cautiously: at other times he did not shrink from imagining an elaborate mechanical model. All physicists know its principal features. The magnetic energy is considered as a true _vis viva_, the magnetic field being the seat of invisible motions, rotations of small particles about the lines of force. The system of these particles may be compared to a wheelwork, and Maxwell has to explain how it can be that all the wheels in an element of volume are rotating in the same direction. This shows that the motion is not transmitted directly from one wheel to the next. So Maxwell is led to assume that between these wheels of the magnetic field and in contact with them, there are smaller ones which transmit the motion in the manner of friction wheels. What I called wheels might in reality be spheres capable of turning about axes in any direction and, as Maxwell showed, a system of this kind is amenable to mathematical analysis. In his image the friction wheels represent what we call electricity; in a conductor we must conceive them to be freely movable, whereas the centres of the larger wheels or balls have fixed positions. It is easily seen how a motion of translation imparted to the friction wheels can give rise to rotations of the larger balls; this is how a current produces a magnetic field. Other phenomena can be explained on the same lines; it is sufficient for this that Maxwell's equations can be deduced by means of the model. Similar models have been invented, with more or less success, by other physicists, in such quantities, even, that nearly all that could be done has been tried. The outcome of these various attempts is, in my opinion, this: that we must admit the possibility of mechanical representations, a possibility that is already shown by the fact that the formulae of the electromagnetic field can be given the form of the general equations of dynamics. On the other hand it cannot be denied that, when we desire them to be applicable to a comparatively wide range of phenomena, the theories that have been proposed become so complicated that they can give us but little satisfaction. So, I think, the majority of physicists will agree to attach less importance to mechanical models and even to the general dynamical analogies, however helpful these may be to us, than to Maxwell's equations that summarize so admirably all that is essential.