A very economic method of organizing records in an information retrieval system is due to Ghosh [1,2] (see also refs. [5,6]): the records relevant to a query are stored in consecutive storage locations, and one location may be significant for several queries. Thus, the record sets corresponding to queries form overlapping segments in storage. Unfortunately, this may in general not be done without repetitions, so redundancy has tobe taken into account (see refs. [3,7]). Putting it into more abstract terms, the problern is as follows: given a family cm_ == {M1,M2, ... ,M11 } of subsets of a finite set X, find an arrangement of all elements of X suchthat (i) each Mi forms a segrnent and (ii) the total number of repetitions is minimal. The problern (for linear arrangements) has been shown tobe NP-complete, see ref. [4]. In i:he present paper we give methods for producing suboptimal solutions, and we calculate the storage space required. Our basic assumptionisthat each component of the given family cm_ contains exactly one element. (A component of cm_ is a block of the product of all partitions {MbX\MJof X, see ref. [8]). This assumption seems tobe not too restrictive, if we consider the method e.g. developed by Marek and Pawlak [8]: each component is stored elsewhere and referred to by a pointer; thus, we have 2n pointers referred to by the 2n binary n-tuples as keys. Our problern now is to find arrangements of these n-tuples such that, for each position i, 1 < i < n, there is a segrnent all of whose items have ith position 1, while the segment contains all possible 2n-l items, each one exactly once. This Situation is somewhat related to the inverted file organization with n binary attributes in the case ofuniformly distributed keys. Here,Ln == n2n-l storage locations are required. The storage space required by our linear method (section 1) will be 2 shown tobe about 3Lw In section 2, the method is generalized to the arrangement of items in acyclic f-graphs as introduced by the second author [5] (see also ref. [6]). This organization will be shown to need approximately ~Ln storage locations.