The problem with which we shall be concerned, in the whole of this Second Part, has already been described. It is the determination of the path of our model economy (the Full Performance or maintainable path) when the economy is not in a steady state. Such a path must have a definite time-reference; for, out of the steady state, one point of time is not like another. In particular, it must have a beginning. The path which follows from that beginning is what we have to determine; so the state of the economy at the beginning (and its previous history, in so far as that is relevant) must be taken as given. One would like to assume that this initial state is itself a mixed state, itself the result of a transition which is still incomplete; but a state of that sort we do not yet understand. So it seems inevitable that we should begin from what we do understand—that we should begin with an economy which is in a steady state, and should proceed to trace out the path which will be followed when the steady state is subjected to some kind of disturbance.
That is why I propose to consider the problem as one of 'Traverse'. We begin with an economy which is in a steady state, under an 'old' technique; then, at time 0, there is an 'invention', the introduction of what, in some respects at least, is a new technology. Among the new techniques which thus become available, there is one which, at the initial rate of wages, is the most profitable; so, for processes started at time 0 (or immediately after time 0) it is adopted. The new technique is adopted for new processes, but the old processes are continued, so long as it is profitable for them to be continued. However, in the course of the adjustment which follows, the rate of wages may change; and, as the result of the change in wages, without any further change in technology, a third technique becomes dominant. New processes then use the third technique, while the first and second (it may be) are still in operation. This is the kind of sequence, involving changes in wages and