ABSTRACT
Up to now, the initial state of the instrument has been ascribed a state vector. This convention can be criticized, however, either by observing that the instrument is a complicated system which is therefore imperfectly known, or, more adequately still (this is again the difference between proper and improper mixtures), by noting that the instrument has interacted with many other systems before the time when it is used for measuring L, and that consequently it can be associated, not with any definite state vector, but only with a density matrix (in other words, an ensemble of such instruments identically prepared would be an improper mixture). We may wonder whether this remark offers a clue for solving the difficulty encountered at the end of Chapter 14. There it was shown that, if the ensemble E of instruments is initially a pure case, it is in general impossible to consider that every instrument coordinate has some definite eigenvalue after the interaction with the system has taken place. Let us investigate whether or not this is still the case when ensemble E is a mixture.
