Using MARC for Collection Management
The Arguments Against

by
Robert A. Baron
Museum Computer Consultant
 

as published in Spectra, vol. 22, no. 4 (Summer 1995)
as an answer to "Help?!" (Vol 22, No. 3)
 

Home Page

QUESTION: Are there museums using the MARC format for cataloging objects? How well does it work? What are the reasons for using it?

ANSWER: I think that the questions on the use of MARC for object cataloging should have included the option of discussing the reasons NOT to use it.

The tagged MARC system employs a data model that was designed to be used as a protocol for transferring data by linear tape or in single files. As such, it is structurally unsuitable to be employed in data models created for relational or entity database systems. For instance, in MARC, specific tags (fields) identify the various maker roles and the kinds of titles associated with a work. On one hand, this means that every possible maker role must be predetermined; on the other, it introduces areas of ambiguity when there is a question about how information is to be catalogued. In practical terms this means that it is impossible or highly unfeasible to phrase a request like the following: "Give me all the works that John Smith has been associated with, in each and every role he may have had." Except for the fact that MARC uses a fairly elaborate sub-fielding system, it behaves much like a flat-file database, opting to define the highest level of distinction among the fields, rather than to find the greatest degree of unity. In contrast, synthesis is the goal of the typical relational data model, as seen in the paradigm of the "roleplayer/role" entity in which roleplayers (names) are paired with one or more roles from a continually evolving set.

The analytical and differential structure of MARC makes it unwieldy and complex for untrained users to use and query. Consequently, library OPACs (online public access catalogues) typically offer only a highly simplified and conflated version of the MARC fields that lie behind each entry. Adding to the difficulty of finding information in MARC formatted data are difficulties in establishing authority control over MARC values; for example, in any single institution's MARC database an author's name may appear in many variants, this due to the bibliographic requirement that the title-page be treated as an authoritative information source -- the cataloguer is required to respect it. Respect for sources, of course, is a laudable attribute. Art object cataloging in relational systems, of course, can suffer the consequences of similar dilemmas: how much variation can be tolerated; how much uniformity must be imposed. Relational systems, however, foster uniformity; MARC systems promote differentiation. Today, any good thesaurus/lexicon query engine placed over MARC or relational records should overcome these problems; but, this kind of utility is generally available only in high-end cataloging systems.

Librarians hold that one of the main reasons for using MARC is that everyone else uses it; it has become a "virtual" standard. While this is true (for better or worse) for the library community, it is not true for the museum community -- and for good reason. For the former, a MARC record for one book will be the substantially similar to the record for every other example of that book. MARC records are generic. It makes good logical sense to apply the work of one cataloguer to the holdings of other collections. With the exception of fine-arts prints, books, and other multiples, museum objects tend to be singular items.

Museums catalogue unique objects, made and existing under unique conditions, bearing unique histories. While published books tend to be defined by the information on their title page or within their covers, museum objects are almost always defined and catalogued as a result of a series of attestations and opinions that are not attributes of the object qua object, but rather of the scholarly and other changing worlds that surround the object. This means that whereas book records tend to be self-referential and stable, museum objects typically point outward to an ever-changing array of opinions, resources, other objects and documents that are collected from and defined by curators, visiting scholars, published and unpublished sources, artists, and owners. Museum objects are defined and understood by virtue of their gestalt.

Integral to museum documentation systems is the collection of usage records for their objects, including those for exhibits, published reproductions, and histories of all kinds, including those for loans, conservation, incidents, and publicity. Such ever expanding internal and external lists need to be interlocked in such a way so that they form a network of associations that can link not just to a single object, but to multiples of objects and activities. A question put to an object such as "what is the attribution history for this object in the Eighteenth Century" is not more pertinent than one which asks: "In the Eighteenth century, which objects were attributed to Rembrandt that today are given to his followers." In other words, MARC recording assumes that the object (book) is central to the database, while museum records must be susceptible to shifts of focus in order to meet the evolving demands of varied users.

A linear cataloguing system such as MARC, even when integrated into a relational schema designed to provide "object management" functions, cannot efficiently handle the diverse associative tasks required by the unpredictable information needs of a user network. Only entity relationship models and their variants have been proven to be efficient tools to establish effective intellectual access under these conditions.

On the other hand, if data is to be transferred from one location to another, from one cataloging environment to another, or sent for a specific predefined purpose that posits a specific set of fields, such as might be required to set up transfers or loans of objects, then a tagged format such as MARC or, better, what is developing under the CIMI project, will be as convenient as a relational system is inconvenient. Similarly, if a database has only a limited specified function or a necessarily constrained set of fields, or if it describes objects (archives, rare books and manuscripts, realia) that have been using MARC successfully, and a body of expertise has been built up on the MARC cataloging concept, then there may be sufficient argument to continue with it.

In general, however, while MARC cataloging represents the height of the articulation of intellectual entities, it is mired in an data model that has been surpassed. Embracing MARC for fine arts cataloging will often create more difficulties than it addresses and may keep its users from migrating to new data models as they are invented.

Home Page | Top