Overview of Library Authority File Record Structure, Format, Content, and Processes

John J. Riemer

Assistant Head, Cataloging Department, University of Georgia Libraries, Athens, Georgia.

Scope

I am going to select from the daily cataloger activity of creating and maintaining authority records some of the aspects of structure, format, content, and process that would most likely be of interest to the systematics community.

In the library community there are two authority files, referred to as the Name and Subject Authority Files. (The machine-readable version of the LC Classification schedules would constitute a third authority file, one that has not yet been made widely available.) The Name Authority File governs the headings for persons, corporate bodies (including conference headings), uniform titles, and those geographic entities capable of authorship. Some of the uniform title authority records consist of a combination of a name plus a title. All the control numbers for these records begin with the letter "n" [Fig. 1].  

The Subject Authority File governs topical subject terms, geographic entities not capable of authorship (like a mountain), names with subject subdivision terms, and terms and names used as subject subdivisions. The control numbers for these records all begin with the letter "s" [Fig. 2]. Guidelines exist for certain ambiguous entities, telling catalogers in which authority file to establish a heading. At various times, LC has spoken of combining the records into a single authority file.  

Structure

Machine-readable authority file records have 3 segments [Fig. 3]:

1. The leader forms the first 24 characters of each authority record and indicates the overall record length, the status (new/corrected/deleted), the degree of completeness, and a signal that the type of record is "authority" instead of some other variety such as "bibliographic."
2. The directory is a computer-generated index to the rest of the record, noting the beginning and ending point of all the other fields.
3. The variable control fields vary in their length, in their required/optional presence, and in their repeatability. These fields are all identified in the directory by a three-digit tag, and they contain either a single data element or a series of fixed-length data elements identified by relative character position.

Typically, the overwhelming majority of the content of an authority record display comes from this last segment.

Format

The variable-length fields commonly found in authority records in both files are these [Fig. 4]:

A control number is assigned to every heading established. Records usually feature an authorized heading valid for use in bibliographic records; some authorities serve solely as reference records, in order to give special guidance for a non-established heading; and other authorities govern just subdivisions to be used with valid headings [Fig. 5]. The 4XX fields indicate invalid terms or synonyms for the heading [Fig. 1]; the 5XX fields provide other valid headings, with which the heading has a relationship, such as earlier/later in time or hierarchically broader/narrower in scope. These 4XX and 5XX references represent one-to-one relationships. The 2XX and 3XX blocks contain fields representing a one-to-many relationship the heading has with other terms [Fig. 5]. The 6XX notes present documentation about how and where on an information object the strings in the heading and reference fields appeared [Fig. 2]. Notes also include directions to the catalogers about treatment decisions and the scope of headings [Fig. 4]. The linking entries (7XX fields and some of the 0XX fields) exist to map headings to terms in other vocabularies and to library classification systems. The 8XX fields can store images of non-Roman-character equivalents.

Content

The control number consists of two digits (usually) representing the year the heading was established, followed by a sequential number within that year [Fig. 2].

For headings in the names file, catalogers follow a principle of calling an entity by what it prefers to call itself, as expressed on the title pages (or other equivalents) in its publications. The substantive portion of headings for persons, the family name and the given name(s), is tied to what appears in these "chief sources of information." Additions, such as terms of address and birth/death dates, are made when readily available with a mind to future conflict-prevention in a very large file. Corporate body names are established in the vernacular whenever that language form can be found; these are not derived by translation, since that potentially arbitrary string may actually never have been used by the body. Conference headings regularly include the conference’s date and venue, to facilitate identification [Fig. 6]. Names initially used only as subject headings are based on the most prevalent or suitable form found in the entire publication. Reference works are occasionally researched for names, when clarification of relationships or facts is needed. Since these tools often contain official forms of name, not necessarily used on publications, they cannot be relied on as a primary source for headings.

For topical subject headings [Fig. 2], catalogers consult standard reference works to learn which term among multiple synonyms currently receives the most widespread usage; making the determination from just a single publication being cataloged would risk an idiosyncratic result. LC’s subject heading system prefers the English or common name for an organism, except when Webster’s Third [New International Dictionary of the English Language, Unabridged] or another standard reference source indicates the term is ambiguous, in which case the Latin term must be used.

Headings in authority records usually refer to an entity in a "direct" manner; they include surrounding context only as necessary [Fig. 7]. To learn an entity’s full hierarchical or genealogical context, one must follow the references throughout a sequence or family of related, individual authority records. For example, in subject headings:  

PHYLUM/DIVISION—CLASS—ORDER—FAMILY—GENUS—SPECIES

or in the case of corporate bodies, all superior levels of hierarchy in descending order, or for personal authors the complete list of pen names under which their works have appeared. Sophisticated programming efforts in the future might be able to build such displays on demand.

References (from invalid strings) are included to cover any variant form of the heading that might reasonably be sought [Fig. 8]. For persons, this extends through the first element to the right of the comma; for the other types of authorities, the references cover variations anywhere in the heading. These variations may stem from forms found on the information object; they may derive from standard permutations of such things as compound surnames, abbreviations, unusual entry elements; or they may represent formerly valid headings formulated according to earlier versions of the rules. Past versions of the heading that were simply in error generally are not retained for historical purposes.    

Notes substantiate the heading choice and document the variant forms that appear in the item being cataloged and in other sources consulted such as reference works, the files of other libraries, and even telephone calls to authors [Figs. 1 & 2]. The absence of a term from a standard reference tool, where it would normally be expected, is noted. Public notes can explicitly define the heading’s scope. Non-public notes can give catalogers additional instructions or indicate the uncertain relationship of the heading to another term.

Process

File Growth

Any name heading not already on file may be added. As appropriate, links to related headings are made to and from the new record. Newly-encountered synonyms are added to existing authority records; the supporting notes added at the same time as these variants both re-substantiate the heading choice and document the new reference. Components of headings common to multiple authority records often work in building-block fashion. The organization in the parent-body position of a heading or reference must be established separately. The contents of many parenthetical qualifiers must also be established for the sake of internal consistency in the file.

At first glance it appears that Lamarck is duplicated several times in the name authority file [Fig. 9]. Control no. n79066729 represents the author alone [Fig. 8]. That record serves as a building block for all the other Lamarck records which exist to establish a heading for individual works by Lamarck which have appeared in multiple manifestations. The record with control no. n85802575 [Fig. 10] represents the English translation of Lamarck’s Hydrogeologie. In the library community, it is considered undesirable merely to present an alphabetical list of titles under the standard heading for Lamarck. It is considered necessary to pull together in a browsable display all the manifestations, etc. of a work with differing title-page titles). Looking at control no. n84017662 as an example, it would be far from optimal to have the original French title "Philosophie zoologique" subarrange quite a distance away from the English translation "Zoological philosophy" in that alphabetized list of titles I mentioned. These two Lamarck authorities created in the 1980s are referred to as uniform-title authority records, since they serve to make uniform the titles under which a given work has appeared. For translations, the title in the original language is used as the uniform title heading, followed by the name of the language, in English, in which the original work has been rendered. (Figures 11-13 demonstrate the same building block progression and accumulation of authority records for Darwin.) The structure and display of a catalog ought to (a) enable a searcher to learn he has a choice of editions and (b) facilitate making a choice.

New subject headings have to be added with a careful eye to how the concept overlaps with existing records. Determining the need to add or modify a name authority record is relatively straightforward by comparison. Some indications that subject heading development work is needed are one's having to resort to an overly broad heading or two to cover a topic (new heading warranted) or having to overcome the lack of a reference to find the appropriate term (modification desirable).

Conflicts

Certain rules govern the headings and references. The string in a heading can never exactly match that of a 4XX synonym field in the same or a different record. That would amount to saying the string is both valid and invalid at the same time; this conflict is resolved by differentiating one of the strings, e.g. by adding a qualifier. The heading in one record can never exactly match that of another. When this occurs with persons or uniform titles, one of the headings is augmented to break the conflict; for corporate, geographic, and subject headings all the conflicting headings receive a qualifier. Authority records representing the same entity are consolidated whenever their existence is discovered (the headings may or may not match). All the pseudonyms of an author (and the person’s real name if that is used in publications) are considered separate bibliographic identities, and they are established on separate authority records. Persons changing their real names are treated as a single entity; this probably makes sense if we have a finite existence. Geographic name changes are treated like personal name changes -- the heading on the single authority record is changed to the new name, so long as the boundaries of the place remain the same. Corporate bodies, with a potentially unlimited existence, receive multiple authority records when their name changes. The 5XX fields, referring searchers from one valid heading to another, are generally made in a symmetric fashion. All changes to authority-record headings are typically accompanied by steps to update the corresponding headings in bibliographic records in files shared at the national level.

If an isolated heading is found not to conform to the rules governing its formulation, any cataloger can change it. Name headings based only on the viewing of a bibliographic record are overturned when they conflict with a first-hand view of a publication. Obvious errors can simply be corrected. When questions/disputes arise with the interpretation of rules and guidelines, these are submitted to senior catalogers with considerable expertise on a Cooperative Cataloging Team in LC’s Regional and Cooperative Cataloging Division.

I have tried to sift out for you the nuts and bolts of library authority work without too much of the nitty-gritty. For further information, I put together a single-page annotated list of additional sources.

Transcript of Discussion

Biographical Information

John Riemer’s responsibilities at the University of Georgia Libraries include Assistant Head of Cataloging, Head of Serials Cataloging, and Coordinator of LC National Programs. For the past decade, he has been Georgia’s NACO liaison to the Library of Congress and its representative on the Operations Committee of CONSER (the Cooperative Online Serials Program). He has promoted participation in a broad range of cooperative cataloging and authority work through writing and experimentation. John holds a B.A. in linguistics and a Master’s in Library Science, both from UCLA.

Additional Information on the Format and Content of Authority Records Used in the Library Community

General Procedures

1. Authority Control: Principles, Applications, and Instructions. Clack, Doris Hargrett. Chicago: American Library Assn. 1990.

   The best single source for a comprehensive look at library authority control procedures

Content

Names/Uniform Titles

2. Anglo-American Cataloguing Rules, 2nd ed., 1988 revision. Ottawa: Canadian Library Assn.; Chicago: American Library Assn., 1988-. Chapters 21-26.

   Rules governing the formulation of headings for persons, places, bodies, uniform titles and some of the guidelines for references to provide. These are augmented in the United States by the following source.

3. Library of Congress Rule Interpretations, 2^nd ed. Washington: Library of Congress, Cataloging Distribution Service, 1989-.

Subjects

4. Subject Cataloging Manual: Subject Headings, 5^th ed. Washington: Library of Congress Cataloging Distribution Service, 1996.

   Roughly equivalent to 2., this source covers full subject heading authority record content as well as heading assignment policies for bibliographic records.

Classification

5. Subject Cataloging Manual: Classification, 1^st ed. Washington: Library of Congress Cataloging Distribution Service, 1992-.

   The LC Classification system equivalent to 4.

Format

6. (A) USMARC Format For Authority Data: Field List, 1993 ed. URL: http://lcweb.loc.gov/marc/authority/ecadlist.html

   A list of fields used in authority records.

7. USMARC Concise Format for Authority Data, 1993 ed. URL: http://lcweb.loc.gov/marc/authority/ecadhome.html

   Descriptions of authority record fields, with some examples.

8. USMARC Format for Authority Data, 1993 ed. Library of Congress, Network Development and MARC Standards Office. Washington: Cataloging Distribution Service, Library of Congress, 1993-.

   The most comprehensive source for this information, with numerous examples.

Items 9-11 represent the same three levels of information about classification record format:

9. USMARC Format for Classification Data: Field List, 1991 ed. URL: http://lcweb.loc.gov/marc/classification/eccdlist.html
   
   
10. USMARC Concise Format for Classification Data, 1990 ed. URL: http://lcweb.loc.gov/marc/classification/eccdhome.html
    
    
11. USMARC Format for Classification Data. Library of Congress, Network Development and MARC Standards Office. Washington: Cataloging Distribution Service, Library of Congress, 1990-.
    
    
12. Descriptive Cataloging Manual. Z1: Name and Series Authority Records. Library of Congress, Cataloging Policy and Support Office. Washington: Library of Congress, 1993-.

    Augmentation of 8, documenting policies followed by LC and its cooperative partners.