Art Libraries Society of North America
Annual Conference, 2005, Houston

“Rules are made to be broken, or stretched: AACR as inspiration”
(Power Point slides)

Why do we want rules in the first place? They allow us to know what to do or not to do. They allow us to interact with our friends and colleagues, as well as strangers. Rules can be very simple or they can be complex. [slide: rules can be simple] In most areas, rules need to be interpreted in order to be applied. This holds for cataloging as it does for religion and politics but we won’t talk too much about the latter here today. There are parallels though. Just as religious commandments are interpreted by churches and preachers, cataloging rules are applied by catalogers and interpreted by agencies such as the Library of Congress or the CCO editors. It may not seem like it to Ann Whiteside and her colleagues, but the rule writing has been the easy part. It will be the applying that is difficult but also stimulating and even enjoyable.

The evolution of our cataloging rules for library collections stretches back to the beginnings of western civilization. Collections of papryi or manuscripts in classical and medieval times had listings of the materials in their collections. Until the mid-nineteenth century, most of this evolution was institution-specific. That is, each collection followed rules that reflected how the librarians and other keepers saw their collection. With the development of Panizzi’s rules for the British Museum Library, adopted in 1839, and Charles Jewett’s for the catalog of the Smithsonian Institution in 1852, cataloging began to move beyond lists of holdings into what we know today as cataloging.

The first edition of Rules for a printed dictionary catalogue by Charles Cutter came out in 1876, the same year the American Library Association was organized. [slide: Cutter] From Cutter’s Rules, we have the well-known objectives and means of leading the user to a book by author, title or subject. And over the last 125 years, we have moved from Cutter to the ALA rules of 1908, 1941, and 1949. The 1949 rules were issued by ALA at the same time that LC issued its own descriptive rules that looked at the principles behind the rules. These principles led through the work of Seymour Lubetzky to the Paris Principles of 1961 which provided the basis for the International Standards for Bibliographic Description (ISBD) and the Anglo-American cataloguing rules. The draft of the first part of AACR3 was distributed to the American cataloging policy makers before ALA Midwinter in January of this year. It’s really more like AACR6 or 7 but the controversy when AACR2 came out in 1978 made us leary of calling earlier revisions of AACR2 a new edition. Hence, we have AACR2R2002 rather than AACR5.

Throughout this evolution, the desire to share cataloging led to the need for standards. You cannot share cataloging without accepting a common way of doing things. And this is where there is so much promise in CCO. As with so much in our lives, automation and telecommunications are important factors in the character and pace of the evolution. [slide: catalog card and slide] When libraries were making their own collection lists and slide rooms were making their own labels, there was little inspiration for development of standards beyond any given institution. Libraries started sharing cards, or mostly using LC’s cards. Slide rooms moved from labels and shelflists to databases. Earlier sharing of information on art works had been through the medium of art historical literature. That is, you and I both used the caption information for a particular slide taken from a book, perhaps expanded by information from Thieme-Becker, Hollstein, Bartsch, or other reference work.

While Cutter included rules for subject headings, the later codes have not included such rules. Subject headings in the United States have rather been under the guidance of the editors of the Library of Congress Subject Headings. Until the 1980s, library catalogers did not have ready access to documentation on the constructing of subject headings. We relied on seeing examples and subdivision instructions in the Red Books or on catalog cards. Since then, the various Subject cataloging manuals, issued by LC, have provided thorough documentation of the rules and procedures for establishing subject headings in the LCSH vocabulary.

We in the art cataloging world have of course also been looking at the Art & Architecture Thesaurus for more than 20 years and watching its development. It however is a thesaurus of terminology. The Guide to indexing and cataloging with AAT, published in 1994 along with the first printed edition of the whole thesaurus, provides significant guidance but there has not been the huge body of examples that one can find in OCLC and RLIN as we earlier had seen them on catalog cards from LC.

[slide: what are you cataloging?] The basic premise of AACR has been to catalog the item in hand, and recent discussions on VRA-L have focused on the question of “what are you cataloging?” One must make this basic decision before you build the cataloging record. It will determine how you answer the questions who? what? when? where? and will also determine how your record fits with other records and how the user gets results from your catalog or database. A favorite example of a sticky wicket in recent VRA discussions has been the slide of a photograph by Erwin Panofsky of a 15th-century manuscript illumination of a Roman sarcophagus. Cataloging rules are easy; it’s the application that can be difficult. But once you decide what you’re cataloging, you can make the follow-on decisions. Part of this decision-making process is your context. In the Panofsky photo archives, I imagine the Panofsky photo would be the focus. In a collection of images of medieval manuscripts where the work being cataloged is the manuscript leaf, Panofsky will become the creator of the image. An important creator, but nonetheless a creator of the image, not of the work. An effective catalog or database will be able to give meaningful results, regardless of the decision.

You may have heard about the Functional requirements for bibliographic records or FRBR which was issued a few years ago by IFLA, the International Federation of Library Associations and Institutions. [slide: FRBR Hamlet] Much of the emphasis in FRBR discussions has been on whether some resource is a work, an expression, a manifestation, or an item. In most examples, there is ambiguity about the levels. What you are cataloging will drive your decisions as you catalog but our catalogs and databases will only be effective if they can combine records in which different decisions were made by different catalogers or from different points of view. [slide: FRBR Palladio] That is, you want the user to find Panofsky and the Roman sarcophagus, or a drawing of Palladio’s Villa Rotunda, regardless of what was chosen as the work to be cataloged.

Cataloging rules as they have evolved in North America and elsewhere have allowed for the significant sharing of records for materials. One hears of libraries, mostly not research libraries, that have no way to get original cataloging done. That is, all of their processing of materials is based on the expectation of finding copy for the book or video or journal or whatever. AACR provides the rules for describing and providing access to the resource. It is normally used in conjunction with MARC to provide a structure for that content, as well as in combination with rules for subject analysis and classification.

There have long been sources for part of the work of cataloging for visual resources. Tansey’s guidelines were used by many institutions. The Fogg classification was used by collections other than those at Harvard. And, in a sense, the descriptive side of cataloging visual resources was dependent on art historical literature. But the purpose of scholarship is not to use controlled vocabulary or to speak in concise fields of data.

For books and serials, the raw material for a cataloging record is often inherent in the material itself. For example, you normally have a title page and its verso, or the colophon, and the count of pages. Mostly, we are building a description of words from words. This is not true for images and makes the work of the cataloger more challenging. The raw material may be present in the caption in a book or in the vendor’s catalog but “what we are cataloging” is visual and that pictorial content must be turned into words.

[slide: AACR vs CCO] Cataloguing cultural objects is spoken of as a data content standard and it is most closely parallel to AACR among the library tools for cataloging. Nonetheless, it goes beyond the basic description and access rules in AACR. The third edition of AACR has a projected section on authorities but we have not yet seen that. CCO has a whole section on authorities as well as talking about using controlled forms throughout. CCO also talks about XML schema and other data structures while AACR does not. Some see this as a significant problem with AACR because most catalogers in the U.S. are using MARC and the distinction of content and structure rules is mostly not productive. That is, though CCO is primarily about data content, the inclusion of structure and other guidelines will mostly be helpful to catalogers.

[slide: control vs display] Similarly, CCO talks about control and display of data. AACR assumes that the description will be mostly displayed as you catalog it. Personal names will be entered as you expect them to display to the user. For visual resources, one can imagine wanting to display a name in direct order, that is, Andy Warhol rather than Warhol, Andy, when you’re displaying that data in a caption on an image for a course or on a wall label in a gallery.

[slide: AACR vs CCO again] The principal parts of AACR2 deal with descriptive rules and with access points. The descriptive rules, which are based on the areas of ISBD, are rooted in format, that is, books, serials, maps, electronic resources. ISBD however is based on areas, such as title, imprint, physical description. CCO is based on the elements or categories of the VRA Core. This eases the cataloging process because catalogers can make all of their decisions on a particular aspect from the same section of rules. In AACR, titles for example will be transcribed according to one rule and a note about the source of that title will be taken from another rule. It was hoped by some that development of AACR after the second edition would be more element-based. Here, the millions of records in our opacs and in OCLC and RLIN provide a challenge in that they are built in MARC, based on the cataloging rules as they evolved over the past century. The basic MARC structure still just really builds a good card though we usually see it in some opac format rather than in paragraphs on a printed 3x5 card. I think our inspiring can work both ways. While CCO may look at AACR as the progenitor of cataloging rules as a basis for shared cataloging, library rule makers can learn from CCO as well.

As I look at CCO with eyes that have used AACR for thirty plus years, I have a number of issues that I’d now like to discuss.

[slide: work type] Work Type is a basic element in cataloging visual resources. It provides a fundamental categorization of what you’re cataloging. Part one of AACR is divided into a general chapter and individual chapters on particular formats. Our determination of format determines how we deal with the resource but we don’t explicitly state a “work type” except in the MARC format. RLIN used this MARC element to divide its files until the recent database migration. OCLC and other databases have long used this as a limit for search results. Visual resources catalogs will, I think, need to think hard about how broad they want Work Type to be. [slide: basilica and parliament buildings] CCO recommends using the singular form of a term if there is one of them, and plural if there are more. For example, “basilica” as Work Type for Saint Peter’s but “parliament buildings” for the Houses of Parliament. While this may feel right, it could present retrieval problems especially if the plural is not just the singular with an “s.” This may however be based in my LCSH- and MARC-ishness without an assumption that the data will be tied to a thesaurus like AAT where the singular and plural forms may both be present.

CCO recommends capitalizing all significant words in a title. This will be hard for me as a book cataloger with AACR capitalization done more in sentence style.

[slide: Bronzino] AACR spends a good deal of time on constructing uniform titles. While CCO talks about using controlled terms and what you might base the terms on, there is not a similar concept of uniform title. Again, the structure plays an important part. In library catalogs, we rely on uniform titles not only to organize the file of works by a particular person or corporate body but also as the text for relationship access points. That is, if we have a book about Plato’s Symposium, we use the uniform title for that work as an added entry. One looks with envy at databases that are more relational. The only area in which most library opac systems even come close is in linked authority records. Even there, it is usually just the authorized heading which makes it onto the individual bibliographic record.

While we are looking at this Bronzino comparison, let’s talk a little more about controlled vocabulary and how it fits into a data structure. Each of these versions of these Bronzinos are actual examples from the LC/NACO Authority File and from CCO. Each of them is dependent on their context. That is, the LC/NAF version of the Bronzino work is built from AACR and the LCRI on named works of art. It structures the heading to fit the MARC structure. The portrait from CCO uses the elements of creator, title, and others to build an identification for this painting. Can you use the LC/NACO authority record in a CCO context? If you do, you will have to separate the artist and uniform title into the appropriate CCO elements. After you have done that, is it any longer adhering to the original authority file? The form of Bronzino’s name is displayed in direct order in CCO but inverted with surname first in LC/NAF. The title is probably the most transferable element when it is a unique title. However, when a repository is added to identify a generic title, there will be significant variation. For this Bronzino example, the title is unique in LC/NAF; if the title were generic as in the CCO example, the repository would probably be added but it would be given in its LC/NAF form, not in the current location elements of repository and place.

I have not totally resolved in my own mind how this conundrum plays out in authority control. What I can accept is that the form of the artist name and of titles can be taken from one context to another. That is, you can accept the same forename and surname elements for a personal name. You can use the same title even though it will be structured in different ways in a MARC or CCO setting. Only considerably more experience with CCO will enable us to determine if that can be considered controlled vocabulary.

At NYU, we’ve recently discovered that one of the vendors for a set of analytic records purposefully uses their own form of author’s names even if they know what is in LC/NAF. This is disturbing but the publisher is following their interest. For example, a name in LC/NAF may include no dates or only include a birth or death date or a particular version of uncertain dates. The publisher -- Alexander Street Press -- will use both dates if appropriate. The reason for what some see as incomplete entries for some people is that once a name is established, it is better for shared cataloging reasons to keep the heading unless a conflict develops or an error is found in the heading. As a NACO participant, I buy this approach even though it means that Andy Warhol is not dead in my opac. The publisher’s analytic records serve as access within their sets as well as being sold to libraries and loaded into opacs. In their setting, the incomplete dates are not helpful. One might argue that the ULAN approach of clustered headings would solve this but library opacs òwill not integrate clustered authorities in the foreseeable future.

In a sense, the whole CCO record for a work is necessary to provide the equivalent of a uniform title. [slide: daguerreotype] For example, a daguerreotype of “A young mother with her daughter” may only be differentiated by including the work type, descriptive title, and country of origin. In many cases, repository and repository number will provide a unique identification but not for this daguerreotype. The Cataloging Advisory Committee has done some very preliminary work on devising uniform titles for unnamed works of art in a MARC and AACR context. The Morgan Library has built records for ancient Near Eastern seals and other unnamed works and I imagine that Maria will be talking about that in a few minutes.

Earlier cataloging rules for library materials also called for not repeating information that had been given in another part of the description. In a visual resources record, the work type and title might repeat some or all of the words, as might title and style or culture or period. This is most common when the creator is unknown and the title is generic. In cataloging books, we do more repeating now than we did before ISBD but it is limited.

The influence of writing and typing cataloging cards is still felt in the latest published version of AACR2. For example, we abbreviate the publisher’s name. That lessened with the latest cumulation wherein the practice of using “The Center” in the imprint was eliminated. [slide: Des Moines] This was only done if the publisher’s name appeared elsewhere in the description but significantly hampered access to publisher information which can be a valuable access point in an opac. Similarly, commercial publishers are now often given in fuller form as catalogers and rulemakers realize the value of searching on publisher information. Since CCO doesn’t have this history of saving typing, such abbreviating is not suggested in the rules.

The Anglo-American cataloguing rules are decidedly written for English-language catalogs though they have been translated into a number of other languages. The resource you are cataloging usually has a title to transcribe so the cataloger doesn’t need to make a decision about the language of title. If a work has variant titles in the same or other languages, we use uniform titles and/or authority work to get from the controlled form to that on the piece being cataloged. CCO goes further in preferring the English-language form of a title unless the non-English form is commonly used by English speakers. “Déjeuner sur l’herbe” and “Les démoiselles d’Avignon” are common examples. On the other hand, it makes perfect sense to use English for generic titles based on types of works in an English-language catalog. Perhaps the emphasis would be better placed on authorities. That is, if your preferred authority used “San Pietro in Vaticano” for St. Peter’s, you would use that form.

[slide: dates control and display] Book imprints are much more likely to be concise than dates on art works so there are limited rules in AACR for dates of issuance. CCO very nicely lays out how to deal with the less precise dates that are common with art works. The division of labor between display and controlled versions of the date are clearly indicated so that one could display “probably late 12th century” but index on “1150-1220.” Of course, this is another area where data content begins to be structured in CCO. We could also use the MARC format to provide a controlled version of dates but this has not been well documented in library cataloging guidelines.

[slide: Shera’s principles] We’ve talked about some similarities and differences between AACR and CCO. No matter how old or new your rules are, it is really a matter of applying them. A literal application by two catalogers will not necessarily lead to the same answer. And there may be no answer that really satisfies the cataloger or the user. You will grow as a cataloger as you apply the rules and might come to different decisions at another time. I’m not sure that these two principles are correctly attributed to Jesse Shera who was dean of the library school at Case Western Reserve when I went there. I picked up the principles somewhere and later heard them attributed to Shera. They are, I think, true. [slide: share] The fundamental thing to remember however is that you have to use some sort of rules and guidelines, and use them consistently in conjunction with controlled vocabularies, if you want to effectively share your records with others.

Sherman Clarke
New York University Libraries

... go to Sherman's art cataloging, etc. page