Relationships in MARC 21 and FRBR
This paper examines the MARC 21 linking entry fields (760-787) in terms of Tom Delsey’s
Functional Analysis of the MARC 21 Bibliographic and Holdings Formats and Functional
Requirements of Bibliographic Records (FRBR) section 5.3 (Other relationships between Group 1
entities) with a view to determining the potential role of linking entry fields in facilitating a
migration to a FRBR record structure.
The paper attempts to map the various relationships described in FRBR section 5.3 to the
corresponding linking entry fields (if any). It is therefore organized according to these FRBR
relationships, though it proceeds in a bottom-up manner (from item to work) rather than the top-
down manner in which section 5.3 is organized.
Background (linking entry fields)
Linking entry fields were introduced to the MARC serials format following the creation of the
International Serials Data System (ISDS), the predecessor of today’s ISSN Network, in 1976.
The fields were intended to provide a method for linking ISDS records, using the ISSN and key
title as the basis of the link. To facilitate this purpose, the structure of the linking entry fields (760-
787) was made to mirror that of field 222 (key title), with subfields $t and $c of fields 760-787
corresponding exactly to subfields $a and $b of field 222 on the related record (with subfield $x
corresponding to subfield $a of field 022 on the related record). This was the prescribed form of
linking note on CONSER records through 1980, and is still the preferred method of linking at the
international level (See ISBD[CR], 7 Note Area, Introductory note). Because North American
cataloging practice sometimes resulted in the creation of new catalog records (e.g., when a main
entry changed) without a corresponding change in the ISSN, control subfields ($w) were added to
the linking entry fields and explicit linking entry notes (field 580) were added to facilitate record-to-
record linking in these circumstances. The explicit note would be made in the form: Continued
by: [key title], [ISSN], issued by: [main entry].
With the adoption of AACR2 in 1981, the prescribed form of linking note on CONSER records
was changed to the catalog entry of the related record. In revising the MARC format to
accommodate this change, MARBI abandoned the strict subfield-level correspondence that had
obtained between data elements in the linking entry field and the linked record in favor of a
structure where subfields in the linking entry fields corresponded to entire fields in the linked
records. As familiarity with AACR2 grew, CONSER practice changed further, in that a subfield $s
was defined to accommodate uniform titles in certain circumstances, and (for cases where the
related record reflected pre-AACR2 cataloging rules), a subfield $b was defined to accommodate
edition statements (in lieu of uniform title qualifiers). While these changes resulted in a catalog in
which linking entry fields could generate visually meaningful (and more user-friendly) notes, they
further complicated the use of the linking entry fields as far as using bibliographic data (rather
than control data) to facilitate machine linking.
While machine linking can be facilitated by the use of the control subfields $x and $w, even these
present problems. As already noted, the ISSN (subfield $x) may legitimately apply to more than
one AACR2 record (e.g., a printed serial and one or more microreproductions). The record
control number (subfield $w), on the other hand, is not specific to any particular standard identifier
scheme, and must be broken into two parts: (1) a parenthetical MARC organization code for the
Specifically, when a uniform title was recorded in field 130 in a related record, it was mapped to subfield $s of a linking
entry field if it represented a translation but to subfield $t if it represented any other type of uniform title. Likewise, the title
proper (245 subfields $a, $n, and $p) was mapped to subfield $t of a linking entry field if no uniform title was present or if
the uniform title differed from the title proper (other than by the addition of qualifying terms); otherwise the title proper was
not recorded in the linking entry field.
scheme to which the number belongs and (2) the number itself, which must be interpreted in the
context of the related scheme (which may itself require interpretation ).
Mapping FRBR relationships to the MARC 21 linking entry fields
Before jumping into the mapping of FRBR relationships to the MARC 21 linking entry fields, it
may be useful to present the relevant section from Tom Delsey’s Functional Analysis of the
MARC 21 Bibliographic and Holdings Formats:
5.1.2 Relationships between works, expressions, manifestations, and items
Work-to-work relationships are reflected in a variety of MARC data elements. Added entry fields (700-730)
may contain data pertaining to a related work. Certain linking entry fields (770 and 772) are defined specifically
to convey work-to-work relationships, while others (760-786), although not specifically defined as work-to-work
relationships, may contain data associated with the related entity at the work level (e.g., in the form of a uniform
title in subfield ‡s). A number of note fields (510, 525, 555, and 556) also contain data reflecting a work-to-work
Expression-to-expression relationships may appear in added entry fields (700-730), and in certain linking entry
fields (765, 767, and 775).
Manifestation-to-manifestation relationships appear in series added entries (440, 800-830), in certain note fields
(530, 533, and 534), and in a number of linking entry fields (760, 762, 773, 774, and 776). Manifestation-to-
manifestation relationships may also appear in added entries (700-730). Aspects of manifestation-to-
manifestation relationships are also reflected in coded form in field 006/008 for serials and in field 007 for maps,
globes, and computer files.
An item-to-item relationship is reflected in field 544.3
Delsey’s analysis identifies places in the MARC 21 bibliographic format where relationship data
may reside. It does not examine the specific nature of the data that can reside in each place or
the extent to which the MARC 21 field and the FRBR relationship may correspond inexactly or
anomalously. Nor does it examine relationships between different FRBR entity types (e.g.,
In terms of FRBR, existing MARC 21 records are created at the manifestation level: each MARC
21 record for a given work describes a separate manifestation of that work.
FRBR, on the other hand, assigns its various relationships to specific (and multiple) levels within
the Group 1 entities:
For example, post-2000 LC control numbers have a different structure from pre-2001 LC control numbers.
Tom Delsey, Functional Analysis of the MARC 21 Bibliographic and Holdings Formats (Washington, D.C.: Network
Development and MARC Standards Office, Library of Congress, 2002- ). Available online at
Two exceptions to this description exist within CONSER practice: (1) Records created for the US Newspaper Program
represent printed newspapers and any corresponding microreproductions (the latter actually described in holdings
records) and (2) Records created for online serials in aggregations represent the serial as provided online from the
publisher or one or more delegated hosting services and as aggregated in various online databases.
… and all the work-to-work relationships
All the work-to-work relationships
The following analysis will treat each of these relationships in turn, though adopting a bottom-up
organization rather than FRBR’s top-down organization.
Item-to-item relationships are deemed outside the scope of this paper, since CONSER (and
serials cataloging generally) does not describe serials at the item level. For the record, FRBR
defines two item-to-item relationships. The first—reconfiguration—applies to situations where
one item has either been physically joined to (typically bound with) or extracted from another or
physically separated into others. The second—reproduction—would apply only in instances
where the reproduction of a given item resulted in a single reproduction that was itself worthy of
At the manifestation level, FRBR defines only the reproduction and alternate relationships. For
the MARC 21 format, this implies that manifestation-level data in linking fields (i.e. record control
numbers in subfield $w) are appropriate only for these two relationships. Other relationships
(e.g., successor) are defined only at the expression and work levels.
FRBR makes clear that although reproduction is typically a manifestation-to-item relationship—a
single copy is reproduced in multiple copies—it is treated as such in FRBR only when it is useful
to identify the specific item reproduced (e.g., reproduction of a particularly valuable copy of a rare
For practical purposes, this analysis treats control numbers as manifestation-level data, though Delsey in his Functional
Analysis defines them as record-level data (record is one of a number of entities included in his Functional Analysis but
absent from FRBR).
book). Otherwise, reproduction is treated as a relationship operating between manifestations. In
FRBR, reproduction includes macro- and microreproductions, reprints, photo-offset reprints,
facsimiles, and mirror sites.
In the MARC 21 format, no linking field is specifically set aside for reproductions, though fields
775 and 776 may contain reproduction data, depending on the nature of the reproduction. Note,
however, that neither field is restricted to reproductions in the FRBR sense, and may contain
other types of FRBR relationship. Because of this, each applicable FRBR class of reproduction is
described separately below, along with its MARC 21 mapping (if any).
Macroreproductions (e.g., photocopies) of entire serials are rare and seldom exist in more than
one copy. In the MARC 21 format, no linking field is defined for macroreproductions.
The MARC 21 format does not define a separate linking field for microreproductions, lumping
them together with “other physical forms” (e.g., CD-ROM, audiocassette, Braille) in field 776.
While the CONSER Editing Guide instructs catalogers to use this field to link from records for
microreproductions to the record for the original publication, this is optional, and catalogers are
explicitly instructed not to make reciprocal links, except to records for preservation microfilm
master negatives (for which the field’s use is mandatory). In records for microreproductions, the
original print publication is identified by the term “Original” in field 776 subfield $c. In records for
print originals, preservation microfilm master negatives are identified by the term “Microfilm” in
field 776 subfield $c. Note that the CONSER Editing Guide allows the recording of record control
numbers for multiple records in a single 776 field in certain circumstances.
Reprints, photo-offset reprints, and facsimiles
The MARC 21 format does not define a separate linking field for reprints, etc., lumping them
together with “other editions” in field 775. Nor does it distinguish between regular print reprints,
photo-offset reprints, and facsimiles. CONSER practice is to record reprint notes in field 580,
with a non-displaying field 775 to link to the record for the original print publication. No reciprocal
link is made on the record for the original publication.
CONSER practice does not describe mirror sites for remote electronic resource separately.
Mirror sites are identified by separate 856 fields on a single record representing the remote
FRBR describes an alternate as an alternate format (vinyl vs. CD, videocassette vs. DVD) or
simultaneously released edition (by different publishers in different countries). These are
extremely rare with serials. Conceivably, a computer file serial might be (or might have been)
issued in different formats for different types of operating system or for different disk drive
capabilities. Such alternates are not separately described in CONSER records.
Using fields 775 and 776 to enable relationships between records for original publications and
their reproductions would be difficult. As noted above, the fields are not used reciprocally: a
record for a microreproduction or reprint is linked to the record for the original, but not vice versa.
This works for serials cataloging, where, especially for older serials, it may have been necessary to combine parts of
several separate originals to produce a single comprehensive reproduction.
For example, when a single AACR2 microform record corresponds to multiple pre-AACR2 print records.
As FRBR points out, however, facsimiles differ from other forms reproductions in that the intent is to preserve not only
the intellectual and/or artistic content, but the look and feel of the original. A facsimile thus addresses a different user
need from other reproductions, and should, in future, be separately identified in catalog records.
Second, each field is used for purposes other than linking to/from reproductions: field 775 is used
primarily to link to/from other editions, including language, regional, and audience-based editions,
and field 776 is used to the publication in any other physical form (CD-ROM, online,
audiocassette, etc.). (To complicate matters slightly, computer files were formerly linked via fields
775 and 787.) Finally, as noted above, a single 776 field may link to multiple records.
While it might be possible to make these two fields work for reproductions, it would not be easy or
straightforward, and data from fields 775/776 would need to be used in conjunction with the form
of item code (008/23) to identify the type of reproduction represented by the bibliographic record.
A simpler technique for identifying reproductions would be via the ISSN (field 022 subfield $a)—
since the ISSN of a reproduction is the same as that of the original publication—and the form of
item code (field 008/23). The ISSN could be used to collocate the various reproductions with the
original item (and each other), while the values in 008/23 would differentiate the various classes
of reproduction—microfilm (a), microfiche (b), micro-opaque (c), and regular (eye-readable) print
reproduction (r)—from the original (blank) and from each other, at the same time collocating
reproductions of a given type (e.g., microfiche). This technique would have two advantages over
the use of fields 775 and 776: (1) the data used would be concise, unambiguous, and
comparatively easy to access, and (2) they would be able to capture any reproductions that might
not be linked to the record for the original in field 776 or field 775.
The FRBR sub-classification of reproductions implies that other values will need to be defined for
the form of item code (008/23), specifically facsimile and alternate format (the latter to be
specified in field 007/00-01). In contrast to reproductions, alternate formats have no originals.
A necessary digression: What is an expression?
From the FRBR description of work it is clear that FRBR does not intend to be prescriptive in its
definition of entities and their boundaries, leaving much of the fine-tuning of these definitions up
to local implementations and acknowledging that the boundaries will vary within the context of a
given culture or national tradition. The distinctions made within FRBR, therefore, should be seen
as restricted in application to that document alone and not intended to influence particular
Current discussions on FRBR—and on the structure of authority records that will be set out more
fully in Functional Requirements of Authority Numbering and Records (FRANAR)—suggest that
both expressions and works will be governed by authority records. This suggests that
relationships that operate at the level of expressions and works will also be expressed in authority
records rather than bibliographic records, though this question has not been addressed at this
stage. Clarification of this point may need to wait until the publication of the draft FRANAR for
worldwide review, expected some time this year.
The expression is the one FRBR entity that has no equivalent term in Anglo-American cataloging.
This is not to say that we do not create authority records for expressions. We do. It’s just that we
call them name/title or uniform title authority records and leave it at that: We don’t indicate
whether the entity represented by the authority record is a work or an expression. The distinction
between expressions and works may become clearer if we take this authority-record-based
In general, an authority record for a work comprises—at most—three major elements:
1) a main entry heading (unless the work is entered under title)
2) a uniform title stem (typically—though not necessarily—based on the title proper)
3) a uniform title qualifier (if needed to differentiate between otherwise identical uniform
FRBR, p. 16. The FRBR Review Group has appointed a Working Group on the Expression Entity to clarify the intent of
the entity by the use of examples.
Any authority record that contains additional elements in the uniform title (e.g., the language of a
translation) represents, de facto, an expression.
For the most part, this answers the question “What is an expression?” For those of us cataloging
under AACR2, an expression is any intellectual or artistic creation that includes in its uniform title
elements to achieve collocation with the uniform titles of other intellectual or artistic creations.
In the present, mainly unreformed, AACR2 chapter 25, expressions are governed by rules 25.2B
(excluding revised editions in the same language as the original), 25.5C (translation, dubbing,
etc.), and many of the special rules for religious and musical works. In general, chapter 25
reflects the principle that expressions are differentiated to the extent required by users of the
material. Users of religious and musical works require a greater degree of differentiation than
users of most other classes of material. This means that the same degree of variation in different
works will not necessarily result in the creation of parallel expression structures. (Variations in an
early text of Hamlet might warrant such treatment, but variations in the text of this paper might
FRBR excludes “aspects of physical form” from its definition of expression. This would mean that
the parenthetical qualifier “(Online)” and its variants would not be part of an expression-level
authority record. However, this qualifier reflects CONSER/ISSN rather than AACR2 practice, and
may be retired in the future if FRBR is strictly applied.
Expression-to-expression and work-to-work relationships
[While expression-to-work relationships are described in FRBR, these are defined as
relationships between a given expression of a work and a totally different work. No examples of
such relationships have been identified among serials. Consequently, they will not be treated
here. All work-to-work relationships—successor, supplement, complement, summarization,
adaptation, transformation, and imitation—are also defined as expression-to-expression
relationships. Consequently, all they are treated under expression-to-expression relationships
The abridgement, revision, translation, and arrangement relationships exist between expressions
of the same work. In these situations, the expression representing the abridgement, revision,
etc., is seen as a modification of the original expression. The successor, supplement,
complement, summarization, adaptation, transformation, and imitation relationships operate
between expressions of different works (or between different works).
Abridgements, condensations, expurgations
While abridgements are not unknown among serials—especially among recurring administrative
documents such as budgets with audiences requiring differing levels of detail—they are rare and
sometimes difficult to distinguish from summaries. One class of serials for which abridgement is
common is foreign periodicals intended for delivery in jurisdictions where censorship is in effect;
however, such expurgated periodicals have not customarily been represented among CONSER
records. Translations of serials may also at times represent abridgements of the original.
In the MARC 21 format, no linking entry field records this relationship.
Revised edition, enlarged edition, state (graphic)
Revision (and enlargement) of an entire serial is relatively rare, though not unknown. In the
MARC 21 format, no linking entry field records this relationship.
FRBR, p. 19.
Literal translation, transcription (music)
Translations of serials may also at times represent abridgements of the original. FRBR notes that
while literal translations are treated as being a relationship between expressions of the same
work, free translations are treated as being a relationship between autonomous works. Serials
typically fall in the former category.
In the MARC 21 format, the translation relationship is recorded in the reciprocal fields 765 and
767. Because CONSER requires that this relationship always be recorded in these fields, the
linking entry fields for translations should operate smoothly under FRBR.
As with the all expression-to-expression and work-to-work relationships, while the ISSN in
subfield $x of fields 765 and 767 would still be valid at the expression and work levels, the record
control numbers in subfield $w would not. Users clicking on a FRBR-based link in a CONSER
record would be taken to an expression- or work-level display of the translation/translated title.
This relationship does not apply to serials.
In the MARC 21 format, succession is a very straightforward relationship, coded in fields 780 and
785 and specified by the value of the second indicator of those fields. The vast majority of serial
linking entries—continuations, splits, unions, and all other conceivable combinations, in full and in
part, including those not yet dreamt of in our philosophy but no doubt waiting round the next dark
corner—represent successor relationships.
Placing the successor relationship at the expression and/or work level has one important
consequence for CONSER records: It breaks the manifestation-to-manifestation connection.
While the ISSN in subfield $x of fields 780 and 785 would still be valid at the expression and work
levels, the record control numbers in subfield $w would not. Users clicking on a FRBR-based link
in a CONSER record would presumably be taken to an expression- or work-level display of the
In the MARC 21 format, the supplement relationship is recorded in the reciprocal fields 770 and
772. It should be noted, however, that this is a more narrow interpretation of supplement than
that provided in FRBR, which includes relationships in which one resource can be said to
supplement another. Because of this, it includes relationships with non-serials as well as serials,
and with resources—indexes, concordances, teacher’s guides, glosses, appendices—other than
those that explicitly identify themselves as supplements.
Fields 770 and 772 would not capture resources that FRBR treats as supplements but CONSER
does not. In some cases (e.g., cumulative indexes) the supplementary material might be
routinely recorded on the parent record. In other cases, it might be cataloged as a monograph,
with an added entry for the resource it supplements.
CONSER practice requires that the supplement relationship be recorded on both the record for
the supplement and that for the supplemented resource. This means that these relationships can
be mapped to FRBR with relative ease. As with the other expression-to-expression and work-to-
work relationships, while the ISSN in subfield $x of fields 770 and 772 would still be valid at the
expression and work levels, the record control numbers in subfield $w would not. Users clicking
on a FRBR-based link in a CONSER record would presumably be taken to an expression- or
work-level display of the supplement/supplemented title.
The examples of this relationship in FRBR are entirely drawn from music, and the FRBR
definition of complement states that “they are intended to be integrated in some way with the
other work, but were not part of the original conception of that prior work.” While there are
complementary relationships demonstrated in the CONSER Editing Guide—typically recorded as
nonspecific relationships in field 787—it is apparent from the FRBR definition that they are not
complements in this sense.
While these relationships are not uncommon among serials, there are no MARC 21 linking entry
fields to accommodate them.
Adaptation, paraphrase, free translation, variation (music)
These relationships do not apply to serials.
Dramatization, novelization, screenplay
These relationships do not apply to serials.
While parodies of popular periodicals and newspapers are not uncommon, they typically
represent a single issue, and would be cataloged as a monograph.
In the MARC 21 format, no linking entry field is defined for this relationship.
FRBR, p. 67.