The business of the structural engineer is to make a design to meet some specified brief – for example, a steel-framed factory to house a manufacturing activity; a bridge to span a wide estuary; a gantry to carry an overhead cable for an electric train. Design criteria must first be identified – heavy crane loads may be critical for the factory, wind-induced vibrations for a suspension bridge, accurate location of the cable for the train. To satisfy these criteria the engineer makes calculations, and it has proved convenient, during the last century and a half, whether explicitly recognized or not, to divide the engineer's activity into two parts.

In the first stage, The Theory of Structures is used in order to determine the way in which a structure actually carries its loads. There are many alternative load paths for a (hyperstatic) structure; one of these will be chosen by the structure, and must be discovered by the engineer. This formulation of the problem seems to imply something more than a dispassionate search for truth; the structure seems somehow to have anthropomorphic qualities, and indeed nineteenth (and twentieth) century notions such as those of ‘least work’ may colour the engineer's judgement. For example, in forming such ideas the designer may assume unthinkingly that of course there is an actual state of the structure in which it will be comfortable. It is, however, a matter of fact that the structural equations are extremely sensitive to very small variations in the information used in any structural analysis, and that the structural action, the state in which a structure finds itself to be comfortable, can show enormous variation caused by trivial imperfections in manufacture or construction.