Bibliographic Standards Committee
Guide to Rare Book Records in Online Systems
An open meeting was held at the ALA Midwinter meeting on Friday, January 9, 1998 at which the following document was discussed. It represents the consensus of the Committee and others who contributed either at the meeting or by correspondence with the Committee. This document is not fixed: it is expected it will be expanded and modified by the Committee as circumstances warrant.
Please send comments and suggestions to the chair of the Bibliographic Standards Committee.
BSC Guide to Rare Book Records in Online Systems
This document is intended to assist special collections librarians in making effective use of USMARC records for rare books in online public access catalogs. It is addressed to librarians who must communicate with systems staff within their own institutions, as well as those who must work directly with vendors. Some background discussion of the following issues, with further references, was published as "Rare Book Records in Online Systems", by Henry Raine and Laura Stalker (Rare Books and Manuscripts Librarianship, vol. 11, no. 2). Readers of this document are encouraged to bear in mind the following points:
1) The document consists of two parts: a brief checklist of desirable features followed by notes and recommendations. Nothing in this document carries the force of an official ALA standard or guideline. However, it does assume adherence to established standards such as the USMARC format, AACR2r, and various documents produced or endorsed by the RBMS Bibliographic Standards Committee.
2) The goal of the creators is to encourage an informed choice among standard solutions. When appropriate, the reader is encouraged to consult Headings for Tomorrow: Public Access Display of Subject Headings (Chicago: American Library Association, 1992), where issues of heading display are discussed more thoroughly and in a broader context than is possible in the present document.
3) Satisfactory access to rare book records is the result of a complex of decisions made at various levels. These include choice of system vendor, choice of optional features, and local cataloging decisions. Many needs may be fulfilled by imaginative and flexible cataloging practice and do not require any special programming or equipment. On occasion, lobbying for a change in cataloging standards may be the most appropriate solution. This guide is intended to complement a sound knowledge of current library practice and active participation in the profession, and not to replace them in any way.
4) Consistency is desirable in itself. The ordering of browse displays, the arrangement of information on the screen, and the principles of indexing should follow the same pattern throughout the catalog unless there is a compelling reason to depart from it.
I. CHECKLIST OF MINIMAL REQUIREMENTS
1. Special Access Points: It should be possible to input, download, index and display USMARC field 655.
2. Place of Publication: It should be possible to input, download, index and display USMARC field 752.
3. Date of Publication: At a minimum, it should be possible to limit search results by publication date. It is desirable to allow direct indexing of publication date (using either the date fixed field(s) or subfield 260$c). It is also desirable to allow sorting of multiple search results by date of publication.
4. Bibliographic Citations: It should be possible to input, download, and display USMARC field 510. It is desirable that this field should be indexed.
5. Local and Copy-Specific Information: It should be possible to input and display local notes, using either USMARC field 500$5 or 590. By means of labeling or some other device, the patron should be able to readily distinguish copy-specific notes from those that apply to all copies.
6. Local and Copy-Specific Added Entries: It should be possible to input, display, and index local added entries, using either USMARC field 79X or one of the USMARC 7XX fields in combination with a subfield to identify the institution. By means of labeling or some other device, the patron should be able to readily distinguish local added entries from those that might be used by any institution.
7. Relator Terms: It should be possible to input, download, and display relator terms in the appropriate subfields of added entries. It is desirable that headings in a multiple search result sort by relator term.
8. Bound-Withs: In systems consisting of bibliographic records to which holdings (or "item") records are linked, it should be possible to attach multiple bibliographic records to a single holdings record. Once one has retrieved the record for a single item in a "Bound-with" volume, it should be possible to pull up the rest of the items in the volume by means of a simple command or menu choice.
9. Special Characters: It should be possible to download, input, and display the full ALA character set. Digraphs should be indexed as two letters.
10. System Limits: It should be possible to download, store, display, and index any record, field, or subfield of any length from the principal bibliographic utilities. The library should have as much flexibility as possible in suppressing bibliographic data in the public display, at both the record and field level.
1. Special Access Points
Assuming that the library has decided to use USMARC field 655 with authorized vocabulary and that it can be input, downloaded, indexed, and displayed, the following points must be considered:
a. Will it be separately indexed or included with the rest of the 6XX block (e.g., Library of Congress subject headings)?
NOTE (a): Ideally, field 655 should be indexed separately in order to maintain the distinction between topic and form. The following is a browse display from a system in which LC subject headings and RBMS genre terms are indexed together:
(The user has searched for the heading "Plays")
|1 Plays||see Drama||1 entry|
|2 Plays||3 entries|
|3 Plays--1825||1 entry|
|4 Plays, Medieval||see Drama, Medieval||1 entry|
In this example, "Plays" is both a cross-reference for the topical subject heading "Drama" and a genre term. "Plays--1825" is a subdivided genre term. "Drama, Medieval" is a topical subject heading.
b. Will it be included in a general keyword index?
c. Will there be USMARC authority records for terms used in field 655?
NOTE (b), (c): Whenever possible, the indexing and display of thesaurus terms should conform to that of the subject headings in the same system. The answers to (b) and (c) should match whatever decisions have been made regarding subject headings. Furthermore, the order and appearance of browse displays, including capitalization, punctuation, etc., should match those chosen for field 650.
d. How will it be labeled in the public display? In an individual record? As a searching option?
NOTE (d): The survey results on which this document is based suggest that "Genre/Form" is the most widely used label for thesaurus terms. However, the Subject Analysis Committee (ALCTS/ALA) has appointed a Subcommittee on Form Headings/Subdivisions Implementation, with the charge of conducting a thorough investigation of the use of USMARC field 655 and the subfield v in bibliographic databases. Since the Subcommittee's work will be focused exclusively on form data, we expect that their conclusions in this area will supersede our preliminary survey, and we anticipate receiving more guidance in this area. Further consideration might also be given to the fact that all the terms once used in the obsolete "physical characteristics" field 755 are now entered in 655, which has thus been expanded beyond its original scope of "genre/form."
e. Since more than one thesaurus may be used with field 655, should the source vocabulary for each term be labeled in the public display?
NOTE (e). According to Headings for Tomorrow, "If headings from more than one subject heading source are arranged together on the screen, clear identification of the source of each heading is necessary to explain to users why material is found under both headings". The following example is given of a potentially confusing screen display:
see Psychology, Clinical
If such situations cannot be avoided by means of separate topic and form indexes and careful choice and editing of headings, then labeling of the source does seem to be indicated. Headings for Tomorrow goes on to say that if "two sources of subject headings use the "same" subject heading, displays such as ... above can be quite confusing to users. Perhaps when the headings are the same, they should be reduced to one line in the display and the attached bibliographic records arranged in one alphabetic sequence" (p. 49).
2. Place of Publication
Assuming that the library chooses to use USMARC field 752 and that it can be input, downloaded, indexed, and displayed:
a. Will field 752 have a separate index? Should this be a field-level index, a subfield-level index, or a word index? How will the headings appear in browse displays (i.e. will they follow rules for subject headings in order of display, capitalization, punctuation, etc.?)
NOTE (a): If it proves to be feasible to create a separate index for place of publication, the library must decide whether to index field 752 at the field level, the subfield level, or the word level. There are potential problems with each approach:
1. If field 752 is indexed at the field level (i.e. starting at the beginning of the field and reading to the right), the following fields cannot be retrieved with a single search, even though they represent the same place:
752 West Germany$dBerlin
2. If field 752 is indexed at the subfield level, a single search may retrieve an ambiguous result. For example, a search on "Cambridge" would retrieve records with the following 752's:
752 United States$bMassachusetts$dCambridge
3. If field 752 is indexed at the word level, a similar ambiguity may result. For example, a search on the word "York" would retrieve records with the following 752's:
752 United States$bNew York$dNew York
Options (2) and (3) can result in successful searches if Boolean or multiple word searches are available to the patron (e.g. "York" may be combined with either "England" or "New".) Even better results might be obtained from the use of proximity searching.
b. Will field 752 be included in a general keyword index?
c. How will it be labeled in the public display? In an individual record? As a searching option?
NOTE (c): According to the USMARC format documentation, field 752 (Added Entry-Hierarchical Place Name) may be used "to give access to a bibliographic record through a hierarchical form of a place name related to a particular attribute (e.g. for newspapers, the name of the community served; for rare books, the place of publication or printing)." Because of the range of situations in which this field may be used, it is advisable to give it a fairly general label, such as "Place" or "Associated place" rather than "Place of publication."
d. Should there be help screens or authority records (with cross-references) to direct readers to the correct form of heading?
NOTE (d): Authority records, with the cross-references they provide, are a possible solution to Case 1 above. (E.g. the patron who searches for "West Germany--Berlin" would be given a cross-reference leading to "Germany--Berlin".) However, cataloging rules and a USMARC authority format for hierarchical place do not exist. Help screens and written documentation are a necessary but less effective means of ensuring comprehensive and intelligible searches of field 752. Even in the absence of authority control, the consistent use of authorized forms is advisable.
3. Date of Publication
Date of publication is recorded in two places in the USMARC record, the date portion of field 008 (the "fixed field") and subfield 260$c. Neither source of data is ideal for gaining access to records by date of publication: the content of the fixed field is limited by various standards and network conventions, while subfield 260$c may include a wide range of data transcribed from the item, including marks of punctuation and letters of the alphabet. Truly comprehensive and accurate retrieval by date would require that normalized forms (i.e. modern Western-style dates recorded in sequences of four Arabic numerals) be embedded in the MARC record according to agreed-upon rules. Otherwise, partial, but still useful, results may be obtained by direct indexing of the fixed field or subfield 260$c. (The RBMS Bibliographic Standards Committee is sponsoring a proposal that will allow catalogers to record uncorrected and corrected dates in the fixed fields. If this change is implemented, using the fixed fields for date access will become more palatable to rare book catalogers.) If subfield 260$c is chosen, the library must also decide which characters in the subfield to index in cases where more than four occur. If direct indexing of dates is not feasible, the capacity to limit search results by date should be a minimum requirement.
Two more questions regarding date information are:
a. Will the date of publication display along with titles in browse displays? Will it be taken from the fixed field or from subfield 260$c?
NOTE (a): The following is a sample browse display in which the title and subfield 260$c are shown (both truncated):
|1 The abdicated prince, or, The adventures of||1690|
|2 Abra-mule, or, love and empire||MDCCLXV. [17|
|3 An abreviate of Hollands deliverance by and||1665|
|4 An abstract of all the statutes made concer||1685|
|5 An abstract of a treatise concerning the pa||Printed, ann|
|6 An abstract, of certain acts of Parliament||1583]|
|7 Abstract of the bill lately presented to th||1788|
|8 An abstract of the proceedings of the Incor||M. DCC. XXXV|
b. If multiple search results are sorted in order of date of publication should the fixed field or subfield 260$c be used for this?
4. Bibliographic Citations
Assuming that the library uses USMARC field 510, with authorized forms, and that it can be input, downloaded, and displayed:
a. Will field 510 be indexed in the local system?
b. Will field 510 have a separate index? Should this be a field-level index, a subfield-level index, or a word index?
NOTE (b): If it proves feasible to create a separate index for bibliographic citation, the library must decide whether to index field 510 at the field level, the subfield level, or the word level. There are potential problems with each approach:
1. If field 510 is indexed at the field level (i.e. starting at the beginning of the field and reading to the right), the patron must know the exact form of the standard citation, especially the main entry of the work, in order to construct a search.
2. If field 510 is indexed at the subfield or word level, the precision of the search may be sacrificed. That is, a search for "Evans" and "150" might yield not only a bibliographic record containing the following field:
but also a bibliographic record containing two 510 fields, such as:
Proximity searching, which is becoming increasingly common in online systems, promises to eliminate the problems with option (2).
Some systems do not index strings of fewer than three characters. In these cases, citations numbered "1" through "99" would be irretrievable.
c. Will field 510 be included in a general keyword index?
d. How will it be labeled in the public display? In an individual record? As a searching option?
NOTE (d): Since bibliographic citations have a distinctive and easily-recognizable form, it is possible that the choice of label is not of critical importance. Most libraries which give field 510 a distinct label (as opposed to the generic "Note") use some variation of "References" or "Citations." In choosing a label, it is important to bear in mind that in the USMARC format, field 510 has a wider range of applications than its use in rare book cataloging, including access to published reviews of the item cataloged and information about publications in which a serial has been indexed or abstracted. Therefore, the label should be dictated by the indicator used. Indicator 4 is normally that used with bibliographic citations; its label should be "References:" (as suggested in USMARC Format for Bibliographic Data) or something similar, such as "Citations:". Other indicators should trigger other labels (such as "Indexed by:" triggered by indicator 1) in order to preserve the distinction between the two types of information stored in this field.
e. Should there be help screens or authority records (with cross-references) to direct readers to the correct form of citation?
NOTE (e): Authority records, with the cross-references they provide, would be a partial solution to Case 1 above. (E.g. the patron who searches for the citation "Pollard & Redgrave" would be given a cross-reference leading to "STC".) However, a MARC authority format for bibliographic citations does not exist. Help screens or a copy of Standard Citations near at hand are the only solutions available at present.
f. If there is a separate index, what should browse displays look like?
NOTE (f): If there is a separate index for field 510, the multiple, or browse, display of citations from a given bibliography should be arranged in ascending numerical order. (See Headings for Tomorrow for discussion of display of numeric data in headings.) For example, the following headings should display thus:
rather than sorting character-by-character (thus stripping punctuation from decimals), as follows:
Bear in mind that multiple sequences will result from terms used for different editions of the same bibliography (as in this made-up examle):
510 Wing (2nd ed.)$cB3554
510 Wing (2nd ed.)$cB3557
510 Wing (CD-ROM, 1996)$cB3555
5. Local and Copy-Specific Information
Local notes are recorded in one of three standard ways:
1. in MARC field 500, with library's USMARC code in $5
2. in MARC field 590
3. in a holdings record (often called an "item record")
(Notes relating to copy-specific variant titles may be recorded in field 246 with a subfield $5.)
Methods (1) and (3) are always copy-specific. Method (2) is defined as copy-specific in RLIN, and as "local" in OCLC, where field 590 may be used for any note deemed inappropriate for inclusion in the master record.
a. What are the pros and cons of each method of recording copy-specific information?
NOTE (a): In deciding among the three methods, librarians should take into account the impact on (1) workflow, (2) information sharing, (3) intelligibility to the user, and (4) system and utility constraints. In general, including copy-specific notes in the bibliographic record (methods 1 and 2) is preferable to putting them in the holdings record (method 3), for the following reasons:
- information recorded in holdings records is not generally shared through the national utilities or other shared databases
- in many systems holdings records do not have note fields long enough to accommodate rare book catalogers' notes
- in some systems, notes in the holdings record do not appear in the public display
- holdings records are less likely than 5XX fields to be included in keyword indexes
- copy-specific notes are often complemented by special access points or added entries, which are found in the bibliographic record
- information contained in holdings records may be more liable to loss in migration from one system to another, especially, perhaps, in cases where notes have not appeared in the public display
The differences between method 1 and method 2 are relatively slight. However, using a special MARC tag (590) for copy-specific information makes it easy to assign a distinct label (e.g. "Local note") to such notes, if this is desired.
In RLIN, where each library's record is displayed in full, there is no practical difference between method 1 and method 2. However, in OCLC field 500$5 is retained and displayed in the master record, while field 590 is not. For this reason, rare book catalogers in OCLC libraries may prefer to use field 500$5.
b. Should copy-specific notes be included in a general keyword index?
c. How should copy-specific notes be labeled in the public display?
d. When there are multiple copies, how is it made clear which notes apply to which copies? Will there be a cataloging solution (e.g. wording of notes) or a systems solution (e.g. linking of 5XX fields to corresponding 7XX fields)?
6. Local and Copy-Specific Added Entries
Local and copy-specific added entries are formatted in one of two standard ways:
1. the cataloger adds a relator term and/or $5NUC symbol to a standard 7XX field
2. 79X fields are used
As with field 590, the 79x fields in OCLC do not appear in the master record and are not indexed in the OCLC database. In RLIN, the 79X fields are visible and indexed. In both utilities, headings tagged 79X are not required to conform to AACR2r.
(a). What are the pros and cons of the two methods of constructing local and copy-specific added entries?
NOTE (a): Method 1 may be chosen for the following reasons:
- data in the standard 7XX fields is more widely shared in the national utilities (this is especially true for OCLC libraries)
- all instances of a given heading are coded alike, which allows for more straightforward collocation and retrieval in the local system (e.g. the heading for Benjamin Franklin will always be tagged 100/700, whether he functions as an author, a former owner, or a printer)
- network standards do not allow libraries to put local (i.e. non-AACR2r) forms of headings in the standard 7XX fields
- the library may wish to use 79X fields to produce extra cards for special files (this is particularly true of libraries cataloging in RLIN, which provides for special file labels in the 79X fields)
- in local systems, the 79x fields can be given labels distinct from those given the standard 7XX (e.g. "Local name")
- (b). How should they be labeled in the public display? In an individual record? As a searching option?
(c). In the case of multiple copies on a single bibliographic record, how does the reader discover to which copy the added entry applies?
NOTE (c): This question should be considered in conjunction with item 5.d. (regarding copy-specific notes).
7. Relator Terms
Use of relator terms (entered in subfield e), rather than relator codes (entered in subfield 4), is recommended by the RBMS Bibliographic Standards Committee (but see also the last paragraph of this section). Assuming that the library uses them, and that they can be input, downloaded, displayed, and used to sort browse displays of headings.
a. Will one be able to limit a search by relator term?
b. Will relator terms be indexed?
NOTES (a, b): It is useful to sort the browse display of a search result by means of relator terms, thus:
Franklin, Benjamin, 1706-1790
Franklin, Benjamin, 1706-1790, former owner
Franklin, Benjamin, 1706-1790, printer
By this means, all instances of the heading for Franklin can be retrieved with a single search, while at the same time the patron interested in Franklin's works does not have to wade through all the items printed by Franklin. Of course, this arrangement implies that the relator terms will be visible in the public display.
In the example given above, the focus is on access by name (whether personal or corporate). Users who are primarily interested in the function represented by the relator term will find it useful to be able to limit searches (of titles, subjects, places, etc.) by functions such as "illustrator", "printer", or "translator". Direct searching of the relator term, in a separate index, would allow libraries to replicate the traditional special collections files such as printer, binder, and provenance files.
In some systems it is possible to enter the codes, which the system displays to the public as terms. This may result in greater precision in searching, by way of codes, especially where a keyword index includes various fields, while at the same time making the record more intelligible for the patron. The drawback is that only staff are likely to be aware of the possibilities for searching.
Given that all of the items in a "bound-with" volume can be retrieved as a group, there should be a device for conveniently, and in one place, recording information (e.g. provenance and binding notes), added entries (e.g. former owners and binders), and special access points (e.g. thesaurus terms for provenance evidence and bindings) that apply to the volume as a whole.
[There is no discussion for sections 9 and 10 of the document.]