Abstract. This paper reorganizes and further develops ihe theory of partial meet contraction which was introduced in a classic paper by Alchourron. Gardenfors. and Makinson. Our purpose is threefold. First, we put the theory in a broader perspective by decomposing it into two layers which can respectively be The basic idea of partial meet contraction is as follows. In order to eliminate a proposition x from a theory A while obeying the constraint of deductive closure and minimizing the loss of information, it is plausible to look at the maximal subsets B of A that fail to imply x. In an earlier paper, Alchourron and Makinson had proved that when A = Cn(A) taking one such B leaves an agent with too many propositions, while taking the intersection of all such B's leaves him with too few. In [1], AGM investigate the idea of taking the intersection of a select set of such B^s. The choice of which B's to take is made with the help of a selection function. A natural question is whether all these selections can be represented as the selections of preferred £Ts, where the preferences between maximally nonimplying subsets of A are independent of the proposition x to be deleted.