My, What Good Access You Have:
Archives and the Big Red Books

Richard Pearce-Moses

Originally presented at the Society of American Archivists Annual Meeting, Indianapolis, 1994

Copyright 1996. Brief excerpts from this paper may be quoted if fully cited. Further use is prohibited without the advanced written permission of the author.

* * * * * * *

I would very much appreciate feedback from readers on this piece. To contact me, see

* * * * * * *

I beg the indulgence of the reader, as this paper was written for oral presentation, rather than publication. In particular, a few ideas are poorly developed or supported. When I realized the text would significantly exceed the time I was allotted to speak, I turned my attention to other sections of the paper. Although I did not present the undeveloped sections in my talk, I've left them in this draft as "kernels" that will be expanded during revision.

Archivists have long recognized the limitations of provenance as an access tool. Provenance works very well as an access tool in government and corporate archives, where the functions of an office suggests the "subjects" of a collection. Provenance works less well in providing access to secondary research value of those collections or to special collections; too often patrons don't recognize the relevance of collectors to their subjects.

In many repositories, the primary subject access tool is the archivist's memory and knowledge. Speaking from my experiences on both sides of the reference desk, there's nothing quiet like the synergistic dialogue that can happen when the patron and archivist connect on a subject during the reference interview.

Unfortunately, that creative synergy may never happen for a wide range of reasons. The patron may ask about a subject the archivist is unfamiliar with. Deadlines and off-days may distract the archivist or patron from lengthy discussion. The worst challenge may be the frailty of memory; the archivist doesn't happen to think of relevant material.

The ultimate limitation of the archivist's knowledge is it's failure of the "Mack Truck Test." Whether the knowledge is unavailable because the archivist best suited to helping a patron is sick that day, because of the untimely death (due to the Mack Truck), or due to some other change in personnel, the archivist is an imperfect access tool.

Archivists need to find some way to translate as much of their intimate knowledge of a collection to a more permanent medium than gray matter. Finding guides organized by provenance are an essential first step, but subject catalogs are an important complement.


Before we tackle the question of subject access, I'd like to begin by asking a basic question, `What is the subject of an archival collection?'

Names as Subjects

Archives traditionally have limited the sense of subject to the names in a collection; a well indexed archive often meant it had a very good name index. Patrons were expected to know who was relevant to their topic, often having to consult the name index several times as they discovered more and more names in their research.

A name index helped support research based on characteristics of the creator--occupation, ethnicity, religion, gender, or the like. "I'm looking for the work of women poets" or "I'm looking for congressional papers." These queries are independent of the subject of the materials; one woman's poetry may cover love and war, and one senator's papers may contain birdwatching records. But names and their related occupations and avocations aren't subjects in the sense I'm getting at.

Topic/Concepts as Subjects

What is the subject of the census records? The census itself (the workers, the reports, the conclusions), the people counted (as numbers, as individuals, as groups), or something else? What is the subject of a collection of family papers beyond the family members? The events that family experienced, the community to which they belonged, the things they did?

One of the axioms of subject analysis of archival collections rests on its distinction from bibliographic materials. Books, papers, articles, and the like tend to be written with a central theme; archival collections, on the other hand, grow out of an activity that may not have a topical focus.

In developing an indexing policy and choosing index terms, I'm less concerned about the subject of the materials--what the records are about, and more concerned about relevant topics of research.

In one instance the work relates to the subject directly; in the other instance, the work reflects the subject only indirectly. I'd like to suggest that the catalog should not limit subject access only to works that pertain directly to the topic; we need to provide our patrons access by those indirect relationships.

Many people may be concerned that this type of access becomes interpretation of the materials. Assigning interpretive headings violates the principle against labelling works. I'll concede that point, but I'll also argue that any analysis is, at some level, interpretation. More over, we interpret and label materials during the appraisal process; what is appraisal but an estimation of a collection's potential value?

Archives and special collections differ fundamentally from general information repositories in the focus of their collections policy. Archives and special collections often collection material for secondary values. A human right organization might collect 1930s German literature, not because of its literary value but as examples of propaganda. A women's studies library might acquire romance novels not for their stories, but for the evidence they provide about sexual attitudes. It seems odd that we would invest so much money in collecting for specific reasons, then fail to point out those reasons in the catalog.

We may be no better predicting potential research value when assigning subject headings than when appraising for acquisition. But, the risks are lower. An unassigned heading is no different from what we have now, and an assigned heading that never helps a patron is little more than a stab in the dark. But, I think we may strike our mark more often than not.

Finally, a caveat on labelling and making judgments in the catalog. The traditional--and respectable--argument against labeling is that the reader should make up his or her own mind as to the character of the material. I'd spin that about a bit and suggest, if the reader is wise enough to discern the character of the material, then that reader is wise enough to discern inappropriate labeling on the part of the cataloger. I don't want to get side tracked by the issue of labeling, but I will confess to being nervous enough about labeling to say that I think all such judgments should be flagged in description; for instance, "Scholar X cites this material as examples of racist literature."


Many repositories created subject catalogs by applying index terms in an unsystematic fashion. They used whatever term or phrase came to mind. Such a collection of terms is often quite useful. But those catalogs often lack the benefits of true thesauri: authority control, cross-references, syndetic relationships, and scope notes.

Providing effective subject access is bit trickier than it sounds. I'd like to begin by outlining some of the problems I've run into in providing subject access before offering some strategies to overcome those problems.

Form of Headings

One problem is that a collection's subjects cannot be easily reduced to a few terms. Rather, the complexity of subjects in a collection is more likely to require several interrelated terms. Boiling that information down to a few key terms is often an exercise in frustration.

The Byron Harvey collection contains a series of baskets. Not just any baskets, but Native American baskets. And not just any Native American baskets, but patterned coil baskets of the Arizona Whiteriver Apache Indians made in the nineteenth century.

I have a choice of assigning headings that combines the elements into a complex heading:

or I can assign separate headings for each discrete concepts; In a manual catalog using a separate heading for each discrete concept, the number of entries for some subjects becomes overwhelming. The practice of pre-coordination--combining the elements into complex headings--aided researchers by breaking up massive amounts of headings into more manageable units. The practice of assigning headings as discrete elements--a strategy of post-coordination--became popular with the rise of automated catalogs; the computer would narrow the results of a search by displaying only those descriptions that included the headings requested.

Pre-coordination and post-coordination both have their strengths and weaknesses. I confess a preference for pre- coordination and a belief that post-coordination does more for reducing the workload of the cataloger than it aids the researcher. Those biases are based on what I perceive to be a distinct benefit of pre-coordinated headings and a particular problem of post-coordinated headings.

Post-coordinated headings have a significant drawback in complex collections; a record describing, for instance, Civil War battles and Native American religions might receive separate headings for Civil War, Battles, Native Americans, and religion. The post-coordinated combinations "Civil War religions" and "Native American battles" are not valid.

Pre-coordination avoids inaccurate combination of the elements of headings. However, I prefer pre-coordinated headings for the reason that they can be browsed. Archival patrons are often "rummaging" through the catalog looking for clues. They're not quite sure what they're looking for, but they'll know it when they see it. The analysis and detail in complex pre-coordinated headings helps give patrons a better list to work with.

Relationships Between Headings

Pre-coordinated headings have their drawbacks; which element of a multi-part heading will the patron look under? Library catalogs have alternated between two solutions to this problem; direct headings, which organize the headings by natural language form, and indirect (or classified) headings, which group headings conceptually under a key term. The vast number of inverted and subdivided headings points to the historical practice of grouping headings conceptually rather than by language, although current practice in an automated environment prefers headings in natural language order (which--in theory--will be collocated by the computer).

One solution is to assign several headings expressing the same notion in several forms: "Indian baskets" and "Baskets--Indian." The other approach is to assign cross-references. Unfortunately, many automated catalogs seem to have lost these niceties; a copy of LCSH next to the terminal just isn't the same.

In addition to basic "see" cross-references to authority forms, a subject catalog can assist patrons by helping them consider how the headings are related to one another. This notion is far from novel. In the days of cards, patrons often found "see-also" cross-references to aid their research. Those references pointed patrons to broader, narrower, and related topics. LCSH has embraced this more complex form of headings by adopting the ANSI standard for thesaurus construction; "see also" has been replaced by BT, NT, and RT. LCSH's implementation of a standard thesaurus structure in a list that evolved over decades is far from perfect, but I think they did an excellent job given their constraints.

Defining Headings

Finally, any list of headings is plagued by the ambiguity of language. What do we mean by "Indian Art?" How does it differ from "Arts, Indian?" What's the difference between "Indians of North America--Attitudes" and "Indians of North America--Public opinion? We need to let patrons know what the headings mean by defining them, especially when a heading is ambiguous or the distinctions between two similar headings is unclear.

Again, this concept is not new. Card catalogs used to include all sorts of notes explaining headings. LCSH includes a precious few scope notes. We need to be sure we use those notes when assigning headings, and that patrons are aware of them to guide their searches.

The Limits of Headings

Ultimately, headings alone can only do so much work. Headings represent an enormous level of abstraction from the materials; seldom will a heading adequately capture the nuance and complexity of the records. Instead, headings should appear in the context of descriptions that expand the meaning of the heading.

The coupling of description with a heading is, for me, the distinction between indexing and cataloging. A catalog record provides information that can aid in the selection of materials. An index entry cannot aid in the selection process; because it's merely a pointer, a patron will not know if the materials are likely to be useful until after they have been consulted.


I hope that at this point we realize that subject access is valuable. Moreover, I hope I've pointed out ways that subject access is more valuable if it's transcends being a mere index by incorporating a subject specialist's expertise through simple and complex cross-references and notes.

But do each of us have to start from scratch? Can't we share the work? In fact, that's precisely why so many archivists use LCSH in spite of its limitations. LCSH benefits archivists by allowing them to concentrate on cataloging rather than reinventing a subject authority list and by allowing us to interfile our descriptions in other catalogs.

LCSH is far from perfect, but it does begin to address some of the problems I've mentioned. I'm not going to indulge in bashing LCSH. No one that I know--including book catalogers--loves LCSH. Headings are formed inconsistently: some are complex concepts grouped conceptually by inverted form or subdivision; others are fairly atomic concepts in direct order and scattered throughout the alphabet. Many headings contain aspects other than subject, such as format and chronology.

LCSH reflects a long and complex history. Every complaint I've just listed reflects changes in philosophy of subject access over the years at LC, and each change was intended as an improvement over an existing problem.

I've long admired the principle Peter set forth when he began the LCSH-AMC discussion list: LCSH is an imperfect tool, but rather than waste energy complaining, how can we make the best of it?

Specificity and Accuracy of Headings

I suspect the major complaint about LCSH concerns the headings it doesn't have, not the headings it has. When cataloging a work on brand inspectors--a fairly important office in the state of Texas--I found no such heading; scope notes prevented me from using somewhat synonymous law enforcement headings, such as detective, police, or the like; instead I had to rely on the principle of using a more general heading--"county officials." Now, no researcher in their right mind is likely to look for brand inspectors under "county officials."

A related problem is one of accuracy and subtle nuances of meaning. Recently I was hunting for a heading for a collection of photographs of Native American baskets. The LCSH heading was "Indians of North America--Basket making." The photographs showed finished baskets, not their manufacture, so the collection was not terribly valuable for studying the making of baskets. Although not an exact heading, I can use the general heading "Baskets."

Archives need greater specificity and more divisions under a topic than more general libraries. Yet, archives need this specificity only in the areas in which they specialize. At the Arizona Collection,(1) we had enormous amounts of Arizoniana, but so much of it wound up under the headings, "Arizona--History" with three main chronological subdivisions. I needed more headings to break up the Arizona materials. But, LCSH served me just fine for the bits of medical literature that crept into the collection.

But the problem is not entirely with LCSH. First, many archivists have little or no training in assigning subject headings, and they're unaware of the Subject Cataloging Manual or the Cataloging Service Bulletin; without that knowledge, use of LCSH is going to be even more frustrating because they won't find constructed or pattern headings. Second, repositories who resist establishing local headings are shirking some of their responsibility; no list can be complete, and archivists must build on existing tools.


My favorite rule in AACR2, and the one I consider to be the most important, is the very first:
0.1 These rules are designed for use in the construction of catalogues and other lists in general libraries of all sizes. They are not intended for specialist and archival libraries, but such libraries are recommended to use the rules as the basis of their cataloguing and to augment their provisions as necessary.
That rule has its corollary in the Subject Cataloging Manual (4th ed., 1991):
The guidelines in this manual apply primarily to assigning headings to books and serials in print forms. Since it was not designed specifically for cataloging non-book materials such as motion pictures, computer files, microfiche, posters, realia, etc., many of the instructions may not be applicable or relevant to those who catalog these types of materials. If special materials catalogers wish to follow the general principles in this manual, some adaptation of the rules will be required.
Those two rules point to what I consider to be an underlying belief of catalogers I admire. The rules are a starting point; they are not an end in themselves. I believe the quality of cataloging is not measured by how closely the rules are followed, but ultimately by whether they aid patrons in the selection of relevant materials.

LCSH is a starting point, not the end-all and be-all of indexing vocabulary. In addition to local headings, each repository needs (must, according to the SCM) to develop guidelines for the application of headings (both LCSH and local). Those local guidelines may differ significantly from the bibliographic traditions because of the organic, rather than topical, nature of archival collections.

For example, at The Heard Museum, we are developing a formal indexing policy that assigns direct headings along three principal access points: topic, cultural group, and location. This policy may result in some overlap between headings and sometimes breaks the prohibition of assigning broader and narrower terms to the same record; for example, Hopi Indians, Kachina dolls (a subdivision of Hopi Indians), and Hopi Reservation.


I hardly consider myself an expert to offer solutions or strategies to improve subject access. To the contrary, I sometimes feel like I'm floundering in a sea of words. But, I'm happy to share some of my faux pas and false starts to save you the trouble of committing them yourself. At the same time, I think I've come up with some interesting approaches which need to be tested.

More Training

First, I suspect that many archivists, like myself, have had no formal or academic training in the use of LCSH. In fact I suspect that most archivists have had no training in the principles of indexing. However, there are any number of good books out there. Archivists have a long tradition of on-the-job training and self-development. Before anything else, the profession needs better training within the profession. If you don't have training, get some!

The Committee on Archival Information Exchange is developing courses to be offered in regional workshops and before the annual conference. Take advantage of LCSH-AMC. Read! Ask questions; don't be afraid you're going to look stupid! What if you ask a really basic question; chances are people will think it's enormously profound of you to examine the fundamental aspects of subject analysis!

Setting Indexing Policy as an Aid to Consistency

Second, once we know something about how to index generally, we need to know how to apply that knowledge locally. If the repository doesn't have policies and procedures for subject analysis, they must be developed. That policy will guide archivists in assigning headings along such a vague notion of "topical relevancy."

At The Heard Museum I'm looking to three sources for guidance in setting indexing policy.

Local Headings

Archives must be willing to develop local, specialized headings to complement LCSH. Notice that I suggested training and policy before the creation of a local list of headings. Local headings should be developed carefully. The list of local headings should include cross-references that integrate those headings into LCSH. The headings in the local list should have scope notes defining them and catalogers' notes guiding their use.

In addition to local headings, repositories should look at writing scope and catalogers' notes for LCSH headings to see that those headings are better used. Ideally, the local list would also include LC headings related to the subjects in the repository. This list would be a crib sheet to complement LCSH; by bringing out headings most likely to be used, catalogers will know which of the thousands of LC headings to be most familiar with.

I very much support an idea suggested at the Airlie House conference on the future of subject subdivisions.(2) Allow the use of free-form chronological subdivisions. Instead of limiting date subdivisions to established headings, subdivide the heading based on the precise dates of the materials. This trick allows more precise and fewer entries under a given heading. For a collection that contains Arizoniana from the 1940s through the 1970s, the single heading

instead of the two headings now necessary Unfortunately, LC has not yet adopted this recommendation.

It may seem that the more we rely on local headings, the less we will conform to the standards of national bibliographic utilities. However, we can maintain consistency by assigning authority headings in 650 and local headings in 690. I fear that strategy ultimately means doing the work twice, and we often don't have time to do it once.

One solution would be to use 650 with the second indicator byte set to '4.' However, not all utilities will permit this for National Level Records.

My second solution would require a change in the MARC format to allow local subdivisions of LC authorized headings in 650 and 651. The local subdivision would be an addition to an authorized heading. The local subdivisions would have distinguishing MARC tags, so that bibliographic utilities could distinguish local subdivisions from authorized subdivisions. National utilities could display only the authority form; local OPACs would display the local form.

The notion of local subject subdivions could be extended to name headings. Because archival collections are formed around an invidual, archives need to be able to subdivide a name heading. Currently, the SCM allows names to be subdivided only by a limited list of free-floating headings. Subdividing the name by an existing LC heading rather than a free-floating heading would provide greater specificity while retaining benefits of a shared authority.

Levels of Analysis

Archivists are interested in being able to locate things. Unfortunately, they often want to point to very different things. In many instances, we're pointing to collections or substantial portions of collections that relate thematically to a patron's query. But every once in a while, the archivist runs into a significant item that's lost among its context; a letter from Marilyn Monroe in a collection of banking records. How do we provide access to these details?

I say, "Catalog it." Archival cataloging has emphasized collection level records that point patrons to the finding guide. I say, the finding guided is merely a compilation of descriptions. Those descriptions should be entered into the catalog and traced.

If all those descriptions are fully indexed, you can imagine the potential nightmare. A search for Carl Hayden retrieves virtually every folder description in his 600 linear foot collection! Results like that are no better than dumping the haystack right on the patron. Instead, we need to ensure that our systems point to the appropriate level of description. Someone interested in Carl Hayden is going to want to see his collection; the description of the collection contains an abstract of the components, so a search on Carl Hayden should retrieve the collection rather than the components. I've accomplished that by setting a policy that a heading is never applied to a description if that heading is applied at a higher level. I trace "Carl Hayden" at the collection level only; I don't use that heading with any of the component descriptions.

That torrid love note from Marilyn Monroe to Charlie Keating? It should not be mentioned or traced in the collection level record. The headings should pertain to all the materials described at that level, and the Monroe letter is only a tiny piece. Instead, describe the letter (or the folder, if that's the lowest level you're indexing) and assign the heading there. A search on Monroe will not retrieve the Keating Collection, leaving the patron wondering why a search for something steamy produced banking records.

My suggestion runs somewhat against archival tradition. An esteemed colleague tends to put all the information into a collection-level scope note, followed by bare-bones lists of folder headings. My approach is to layer the descriptions, distributing the information more evenly throughout. Instead of an extensive collection level note (which run for screens and, as a result, are never read), I emphasize more description at the component level by writing a scope note for series and folder headings. Of course, most folder headings do not received a scope note as they are merely instances that are perfectly described by the scope note for the series; but when something in a folder needs to be brought out, it is done at that level.

The Canadians' principles of descriptions require that no component be described before the whole.(3) I admire the principle, but as a matter of practicality, I don't see any great harm in doing in-analytics for something really valuable while skipping the intermediary levels. I'm suggesting, again, that we rely on appraisal to guide indexing practice.

Harnessing the Power of the Automated Catalog

We might be able to take advantage of OPACS to generate complex lists of headings. Rather than listing the headings for a description separately, list them as though they were a pre- coordinated strong. This effectively subdivides each heading by the other headings associated with the records. Browsing headings for a collection that received the headings: would not display just the single headings, but the heading coupled with the other headings for the collection. Then imagine how browsing each heading would be refined by a complex list of all headings.

Finally, we need to investigate expert systems that can help a patron formulate their queries in terms of the indexing language. No only could such a system help improve retrieval by helping patrons select the correct terms to search, it might very well help patrons by guiding them through a process that refines the underlying ideas of their search.


As archivists, we rely heavily on the context provided by provenance and original order to locate items. But our lives aren't so neatly packaged; chaos creeps into the neat order we desire. We need to be able to provide access to the valuable parts of that are out of context . . . . Topical access can supplement contextual access.

I'm suggesting that we quit using word processors to produce finding guides. We need to enter the whole descriptions in databases that can manipulate the information in a variety of ways. Let's take advantage of the computer to sort things to produce finding guides, cross-collection catalogs, and more.


1.  In the Department of Archives and Manuscripts, Univerisity Libraries, Arizona State University (Tempe).

2.  Martha O'Hara Conway, ed. The Future of Subdivisions in the Library of Congress Subject Headings System : Report from the Subject Subdivisions Conference Sponsored by the Library of Congress, May 9-12, 1991 (Washington, D.C.: The Library of Congress Cataloging Distribution Service, 1992).

3.  Planning Committee on Descriptive Standards, Rules for Archival Description (Ottawa : Bureau of Canadian Archivists, 1990).