Archives enable the narration of history. Corporate data centers support decision-making. Libraries deal with ideas and knowledge. Museums and art collections transmit cultural heritage. These outcomes are intellectual and more of less intangible. But the means which enable these outcomes involve very large numbers of highly tangible physical objects, bearing texts, images, and other sorts of potentially informative data. These agencies are, in large measure, collections of data, documents, images, and similar materials. Hence the skillful marshaling of these myriad objects is basic to effective performance and worthy of careful examination.

INTRODUCTION

Physical Medium

Each physical medium-paper, microform, and machine-readable memory-has its own particular characteristics. In particular, it is convenient to cluster them into two categories: those that are "localized" in the sense that the reader and the medium of storage have to be in the same place, as with books on paper or on microform; and those that are not localized and can be read remotely, as with machine-readable electronic databases. Currently, large quantities of materials in archives, libraries, museums, and offices library collections are on the former, "localized" media and all such agencies are much concerned digital records to replace, to represent, and/or to provide access to the non-digital material in their collections. Effective service will, therefore, depend in most cases upon our ability to handle both sorts of media cost-effectively.

Scope and purposes of collections

The scope and purposes of collection, hence the definition of collection development policies, the practice of selecting objects for the inclusion in a collection, and the techniques for evaluating individual collections are highly situation-related, and have had much written concerning them. Here, instead, we are interested in role of collecting per se. If collecting does not create new materials, but with moving copies from one place to another, why is it done? One might say that assembling collections of copies of materials facilitates access to and use of those materials, but this is a vague answer that invites further questions. In what ways do collections facilitate access to and use of materials? If the assembling of collections of copies of materials facilitates access, how might one begin to relate and compare this activity with other uses of these agencies' budgets and space that also facilitate access, such as human intermediaries, help desks, guides, and indexes? How might the effectiveness of collecting be situation-related, dependant, perhaps, on the who is to be served and why, on the nature of the materials collected, and on the service goals of the individual agency?

There is a strong economic motivation for the careful analysis of the roles of collections because the assembling of materials in anticipation of use--as opposed to the actual use of the materials--accounts for a great deal of these agencies' budgets. The assembling for use includes not only the purchase costs of materials acquired but also the labor and incidental costs of selecting, acquiring, describing, and processing the material. For the academic libraries, for example, this preparation of materials for possible use accounts for two-thirds of the libraries' operating budgets. Further, the collecting of materials accounts for much of libraries' space needs, both the space occupied by the collected materials and also the space occupied by the staff employed to select, acquire, and prepare materials for use. It has been estimated again accounts for at least two-thirds of libraries' space needs in the case of the libraries of the nine campuses of the University of California.

The magnitude of the investment being made indicates the importance of clarity in understanding the rationale and effectiveness of this investment.

MATERIAL, COLLECTION, AND COLLECTION DEVELOPMENT

Terminology.

Fundamental to the discussion of materials, collections, and collection development is the distinction between substance (materials) and arrangement (collection). The materials (books, data, images, objects,...) that the people served may need are candidates for being added to or eventually discarded from collections: the selecting and arrangement of copies of materials as a collection is a matter of organization. This distinction is easily overlooked given the ambiguity of the word "collection" which as a common noun denotes a physical entity (a set of assembled materials), but as a verbal noun denotes a process (an assembling).

The terminology of computing is more helpful. The terms "data", "record", and "file" (or, more generally, "files", or "objects") correspond closely to use of the term "material" in librarianship. They denote information as stuff: physically recorded bits constitutive of numbers, text, images, rules, or sounds. The term "file organization" is used to denote how and where files are stored and organized. File organization is essentially concerned with logistics, with how individual files should be arranged, where files should be physically stored, and where and when additional copies should be deployed.

Optimal file organization is highly situational, depending on the number and sizes of files, the frequency, urgency, and nature of the use of each file, the costs and capacities of different forms of physical storage, the speed and costs of transporting files between stores, and the standards of service to be provided. A change in any of these factors is liable to make any given file organization cease to be optimal.

Note that file organization is not directly concerned with the substance or meaning of the contents of a file. The relationship is, at most, indirect. The substantial content of the file is likely to affect the manner in which the file is used, and the nature, especially the frequency, of the usage is a factor in determining optimal file organization. But the relationship is an indirect one.

There is little cause for confusion between "files" (stuff) and "file organization" (deployment). In the terminology of librarianship "material," usually in the plural form of "library materials", corresponds closely to the use of "files" in computing. The difficulty lies in the lack of a clearly understood and accepted term corresponding to "file organization" to denote the "deployment" of library materials, especially decisions whether or not to place copies of materials in particular locations. One could use the term "collection" (as a verbal noun indicating a process) as the analog of file organization for library materials, but it would seem cleaner to use "collection development" or "collection management," or some term clearly denoting a process, the whole process of forming and managing collections of materials.

Collection Development

Collection development has to do with determining the location of copies of materials. There has been a tendency, now diminishing, to use the term to refer only to the selection of materials for acquisition. Adding material to a collection affects where copies may be found, but does not create any new materials. At least there is no direct connection. It could be argued that there is an indirect connection in the sense that some books would not be published, maybe not even written, if there were not a library market for publishers to sell copies to. Books have existence whether or not they are added to any given library collection. The would-be reader of a book can generally be presumed to be indifferent to whether a particular copy of a book belongs to a library, any library, or not. Nevertheless, the would-be reader is likely to be interested in where copies are located. One uses a book from a library's collection because one does not have a personal copy. One's own personal copy is preferable, at least for the period of one's need to use it. In the absence of a personal copy, the presence of a copy in a local library is likely to reduce greatly the inconvenience of achieving access to a copy of the text. It is the effort needed to inspect a copy of a text that is enormously reduced by the addition of a copy to one's local library collection. The essence of a library collection and of collection development, therefore, does not have to do with whether a book exists, but with whether or not a copy is located in a given collection and with the facilitation of access to it. More abstractly, library collections can be viewed in terms of set theory as subsets of changing membership drawn from the broader set of potentially collectible materials in order to achieve the goals of the library by facilitating access by the population to be served.

The development of collections, then, is essentially concerned with the placing in libraries of copies of pre-existing materials. It is, at root, a logistical exercise to improve service. It is this activity of marshalling or deploying that should be the focus of attention in considering both the roles of collections and the scope of collection development.

Any theory of collections would be seriously incomplete if it were applicable to only one type of material in only one sort of context. Military tactics suitable for one sort of soldier in one sort of terrain could be disastrously inappropriate if applied in other situations involving soldiers with different attributes (e.g. differently trained and differently equipped). Similarly any theory of collections or collection development would be seriously incomplete and potentially misleading unless it were sufficiently generalized or flexible to be applied to different contexts and materials with differing attributes.

Goodness

How good a collection is and how well collection development has been effected cannot be considered meaningfully except in relation to some definition of goodness. This can be a measure of quality ("How good is it?"), which, for collections, would appear to be a measure of the capability or capacity of the collection with respect to meeting some definition of intended use. Alternatively, it can be a measure of value ("What good does it do?"), which, for collections, implies utilization and beneficial effects. The definition of goodness to be applied will depend on the goals of the agency's service.

Summary

Any useful theory of collections or of collection development would need to maintain a distinction between the "development of collections" (a process for mediating access) and "materials" (sources which may or may not be collected and whose substance is unaffected by being collected). The theory should be independent of the media collected in the sense of being capable of generalization to media with different attributes and should also be helpful in yielding insight into the problems and opportunities in improving services, including achieving a balance between competing uses of funds, staffing, technology, and space.

THE ROLES OF COLLECTIONS

Collections perform at least four quite different roles:

The Archival Function of Collections

Unless a datum or document is preserved somewhere, it will be lost both in the present and in posterity. The archival role of collections is most clearly seen in manuscript collections, local history museums, scientific data, and specialized collections of all sorts. Even popular and common material is liable to be lost unless it is both collected and archived in collections where it would be available. Essentially it is a matter of preserving selected non-renewable resources in the light of relevant goals.

The Dispensing Role of Collections.

Although information service providers are conscious of the need to preserve copies of materials for posterity, the principal role of most collections, especially library collections, is, in practice, not archival but the need to provide convenient physical access to the materials likely to be needed by the population to be served.

Desired material not held in one's local collection is made available only with delay and inconvenience for the would-be reader and effort. Consequently, other things being equal, the greater the amount of material that is made available locally the better the service. In this situation, agencies should be evaluated by the size and appropriateness of their collections.

The magnitude of the difference between the archival and the dispensing roles is difficult to measure, but, overall, it can be imagined. It is extreme in the case of libraries. Suppose that just three copies of each known text were retained as designated archival copies in the libraries of the United Kingdom and the United States and that all other copies were to disappear. The reduction in the size of libraries' collections would be dramatic. Estimation is difficult, but perhaps the reduction would be by 95%. (Buckland 1975, Potter 1982). The difference in size between the actual and the imagined sizes of collections is an indication of the difference in cost between the dispensing and the archival roles of collections.

The Advisory Role (or Bibliographic Role)

The arrangement of materials for display--on a screen, in a case, on shelves--also permits the collection to play a advisory role in the sense of being a tool for the identification of materials that fit some describable need for appearance, authorship, subject content (Wilson 1968). If the material is arranged by topic, for example, users seeking to identify material on a given topic, in a library, as in a bookshop, can and do seek out the appropriate section and examine the shelves to identify suitable material. This is commonly done as well as, or in preference to, consultation of the finding tool, such as an index. In this sense the array of material displayed is performing an indicative or advisory role analogous to that of a catalogue and quite different from the dispensing role of providing physical access.

Symbolic role

In some cases, most obviously libraries and museums, one cannot explain behavior without invoking the prestige factor of having impressive holdings.

Complexity and the separation of roles

In a simple situation, these three roles are tightly connected. Imagine a single, small, uncataloged collection of a few titles in an isolated location. What has been preserved is what is available and what is available can easily be identified but only by examining the shelves.

In library systems, as in other systems, increasing size and complexity lead to specialization of functions. As complexity is added to library systems, these three roles diverge and become less tightly coupled:

(i) As soon as two or more collections of materials become available, the archival role of any one collection ceases to determine what is available. The user of one collection can extend its resources by drawing on the resources of other collections. Cooperative collection development is a formalization of this opportunity for increasing service through specialization in archival activities and, thereby, extending the overall dispensing capability. The relationship between unique and duplicative items (titles and copies, types and tokens), within or across collections, defines the extent of the preservation role. For any given total set of collections, if there are many copies of an item, the preservation task is relatively small. If all collected objects are unique, with only one copy of each, then the task would be greater. In a digital environment the deployment of additional copies to reduce access time (latency) is trivial, but the different problem of version control becomes more significant.

In an extreme case a library that acquired material but eventually discarded it, referred users to other libraries' collections, and used interlibrary loans could provide extensive library service without any intention of preserving any material permanently. A small public library, for example, may retain indefinitely historical and literary material of local interest, might play a limited and highly specialized role in a "last copy" preservation scheme, and keep some material perceived as special or valuable, yet these three categories might constitute relatively little of its stock at any given time. Industrial special libraries in a fast-changing area can rationally depend heavily on interlibrary loan for infrequent access to less-used material. This pattern becomes more marked as the medium of documents shifts form paper to digital.

(ii) Even a very small number of objects becomes easier to use if arranged in a systematic manner. As the amount of material, the quantity of use, and the sophistication of service increase the limitations of depending solely on examination of the shelves for bibliographic access become more serious. The customary response for non-digital collections in a library is to augment the shelf arrangement by systems of additional, symbolic shelf arrangements using surrogates for the books themselves: catalogue records arranged in various ways. As will be shown below, the use of multiple catalogue records for each book has some significant advantages. A rather extreme development of this augmentation of the advisory role of shelved materials is to depend entirely on catalogue records for this purpose, as can be seen in research libraries with closed access, where users are prohibited from examining the shelves and where the books arranged by size and accession number instead of by subject. (For a good introduction to the theory and practice of arranging library materials see Hyman, R. J. (1982)).

As one considers two or more collections, the utility of a single collection as a bibliographic guide diminishes, depending on the relationship between titles and copies across collections. If each collection were identical, any one collection would be a guide to all that is available. As the collections become more disparate, with less overlap, each individual collection would become less effective as a guide to what exists. (So, too, is the catalogue of any one collection). The traditional response is a consolidated listing, incorporating the contents of multiple collections, a "union" catalogue.

As these examples indicate, a given collection could play a dispensing role without an archival role or without a advisory role, if there has been enough specialization in the broader system of which it is a part.

Bibliographies, catalogues, and the advisory role of collections

The advisory role of collections is one for which bibliographies and catalogues are created and so comparisons can be made between the performance of catalogs and the performance of the collection itself in this role.

Limited coverage: Bibliographies are potentially unlimited in their coverage. Catalogues are limited to what has been acquired and catalogued; collections are limited not merely to what has been acquired and processed, but also to what happens to be on the shelf at the time.

Collection bias. Not only is the array available for inspection limited to those collected titles that happen to be present on the shelves, but the very uneven pattern of demand causes that array to be systematically biassed toward the least used, the least popular, and the least recommended. In an academic library some of the most sought-after titles are usually removed to a reserve collection and those that are not taken are likely to be out on loan if popular. The result is a significant bias in what is on the shelf towards less-used material.

Dimensions of access. A card catalogue ordinarily has two, three, or four dimensions: sequences by author, by subject heading, by classification (classified catalogue or shelflist), and/or by title. Other dimensions such as date, language, and keyword may be found in online catalogues. Bibliographies can be arranged in a variety of ways. The array of material on the shelves is usually arranged in the single dimension of the shelf order adopted, commonly the Library of Congress Classification or the Dewey Decimal Classification.

Single entry. In the array of books on a shelf, a title is ordinarily assigned to one single position. Catalogues and bibliographies can have two or more entries in each sequence, especially in the author and subject heading sequences, either through added entries or through cross references.

Proximity to the text. In each of the four preceding aspects collections compare unfavorably with catalogues and bibliographies as a device for bibliographic access. Nevertheless, it is uniquely attractive because the text is immediately at hand, as it is not with catalogues and bibliographies, which contain merely a brief bibliographic surrogate for the material itself. (It could also be argued that the external appearance of the material provides some visual cues not present in catalogues or bibliographies concerning the nature of the material).

Collecting as a first stage of retrieval

More generally, collecting should be seen as a first stage or pre-condition of retrieval. Material is not catalogued unless it has first been collected. Catalogues are, by definition, ordered lists of collections and are logically and practically subsequent to collection development. It is, therefore, just as well that bibliographies and union catalogues can provide bibliographic access to material not in one's local collection.

Selection decisions determine the set that could be present and loan and duplication decisions affect the subset that is actually present. Acquiring a title because knowledge of its existence is deemed important for users--for "balance", perhaps--is to exercise the advisory role of the collection and is different from selection arguments based on the archival or dispensing roles. This advisory role would appear to be the principal reason that library collections are arranged by subject.