Obtaining and keeping a satisfactory closed beam orbit is a major problem in the design of large accelerators. The system determining this orbit has many variables, these variables are distributed around the accelerator structure, and the existence of the beam is periodic. The designer may attempt to ensure satisfactory values for these variables by brute-force techniques such as ultra-precision construction, or he may apply the more subtle techniques of system control theory. In any case, he is dealing with a multivariable, interacting, sampled-data control problem. For the accelerators of the future, it may not be obvious that a solution exists. In fact, however, there are techniques whereby an equilibrium orbit may first be obtained and thereafter maintained in the presence of all expected disturbances, by means of an electro-mechanical control system employing beam-position monitors. These techniques further permit a wide variety of operator experiments, such as trying various orbits, or restricting the spatial harmonic content of magnet re-positioning. The purpose of this paper is to illustrate certain of these techniques. For clarity we will restrict our discussion to an elementary magnet-position control system for maintaining a closed orbit in the presence of magnet support settling. We assume the beam to be continously present, rather than to be discrete with time. This and other errors, such as magnet imperfections, and other control variables, such as backleg windings, would be handled in the same general way in a realistic design.