Every time you search a library catalogue and instantly pull up a book’s title, author, and shelf location, a structured data record is doing the heavy lifting behind the scenes. That record is most likely built on MARC 21 (Machine-Readable Cataloguing), the standard that lets computers store and exchange bibliographic information. But a MARC 21 record is not just a flat list of details. It follows a precise three-part architecture that tells the system what the record is, where each piece of data sits, and what that data actually says. Understanding this structure is the foundation of practical cataloguing, so let’s break it down piece by piece.
Table of Contents
- Why MARC 21 has a three-part structure
- The leader: the record’s control panel
- What the 24 positions hold
- The directory: the record’s map
- Inside a directory entry
- The data content: where bibliographic details live
- Variable control fields
- Variable data fields, indicators, and subfields
- How the three parts work together
Why MARC 21 has a three-part structure
A MARC 21 record is divided into three main components: the Leader, the Directory, and the Variable Fields (often called the data content). This is not an arbitrary design choice. MARC 21 is an implementation of two formal standards, the American national standard ANSI/NISO Z39.2 and the international ISO 2709, which define a generalised structure for exchanging records that describe all kinds of library materials.
Think of the three components as having clearly separated jobs. The Leader carries coded information the system needs to process the record. The Directory acts as an index that points to where each field begins. The Variable Fields hold the real bibliographic content, such as the title, author, and subject. The standards describe the structure but leave the meaning of the content to cataloguing rules like RDA or AACR2. This separation of “structure” from “content” is what makes MARC records both machine-readable and flexible across different library systems.
The leader: the record’s control panel
The Leader is the first field of every MARC record, and it is fixed in length at exactly 24 character positions, numbered 00 to 23. It is small, but it does crucial work. The Leader contains numbers and coded values that tell the automation system how to process the rest of the record. Each element in the Leader is identified by its relative character position rather than by a tag, which is why position matters so much here.
What the 24 positions hold
Several of these positions are worth knowing because they describe the record’s basic identity and behaviour. According to the Library of Congress concise format for the Leader, the key positions include:
Positions 00-04 (Record length): A computer-generated five-character number equal to the length of the entire record, including the Leader itself and the record terminator. It is right-justified and padded with zeros.
Position 05 (Record status): A single code showing the record’s relationship to a file, such as n for a new record, c for a corrected or revised record, and d for a deleted record.
Position 06 (Type of record): A code defining what kind of material the record describes, for example language material, sound recording, or cartographic material.
Position 07 (Bibliographic level): Indicates whether the record is for a monograph, a serial, a collection, or a component part.
Position 09 (Character coding scheme): A blank indicates the MARC-8 character set, while a indicates UCS/Unicode. This matters for handling non-English scripts.
Positions 10 and 11 (Indicator count and subfield code count): In MARC 21, both are always set to 2, meaning every variable data field uses two indicator positions and each subfield code is two characters long.
Positions 12-16 (Base address of data): A five-character number that marks where the first variable field starts. It equals the combined length of the Leader and the Directory, including the field terminator at the end of the Directory.
Position 17 (Encoding level): Shows how complete the record is, from a full level created by inspecting the physical item to partial or minimal levels.
Position 18 (Descriptive cataloguing form): Identifies which descriptive convention the record follows.
Positions 20-23 (Entry map): These four positions describe the structure of the Directory entries that follow, and in MARC 21 records they are always set to 4500. They tell the system that each Directory entry uses four characters for field length and five for the starting position.
Because every position has a defined job, a single wrong character in the Leader can cause a system to misread the entire record. This is why cataloguers treat the Leader with care even though most modern software fills in many positions automatically.
The directory: the record’s map
Right after the Leader comes the Directory, which begins at character position 24. If the Leader is the control panel, the Directory is the map. It does not contain any bibliographic data itself. Instead, it is a computer-generated index that tells the system exactly where to find each field in the data content section.
Inside a directory entry
The Directory is made up of a series of fixed-length entries, and each entry is exactly 12 character positions long. Each entry contains three parts: a tag, the field length, and the starting character position.
Positions 00-02 (Tag): A three-character code, usually numeric, that identifies the associated variable field. For example, the tag 245 points to the title field.
Positions 03-06 (Field length): A four-character number giving the length of that field, including its indicators, subfield codes, data, and field terminator.
Positions 07-11 (Starting character position): A five-character number showing where that field begins, counted from the base address of data recorded in the Leader.
The entries are not random in order. Directory entries for variable control fields appear first, arranged by tag in increasing numerical order, followed by entries for the data fields. Importantly, the order of entries in the Directory does not have to match the physical order in which the fields are stored. The system reads the Directory, jumps to the right starting position, reads exactly the stated number of characters, and retrieves the field. This is what makes catalogue searches feel instant.
The data content: where bibliographic details live
The third component is the data content, made up of the variable fields. This is where the actual description of the item lives, including its title, author, edition, publisher, and subject headings. The variable fields follow the Leader and the Directory in the record. MARC 21 defines two types of variable fields: control fields and data fields.
Variable control fields
Variable control fields are the fields with tags beginning in two zeros, that is tags 001 through 009, often written as the 00X fields. These fields are structurally simpler. They contain either a single data element or a series of fixed-length coded values, and they end with a field terminator. Crucially, they have no indicators and no subfield codes. Common examples include field 001 (the control number, which uniquely identifies the record), field 005 (the date and time of the latest transaction), and field 008 (a string of fixed-length codes describing aspects like language and date of publication).
Variable data fields, indicators, and subfields
Variable data fields carry the bulk of the descriptive information. These are fields with tags that do not begin with two zeros, such as 100 for a personal author, 245 for the title statement, 260 or 264 for publication details, and 650 for a topical subject. Unlike control fields, each data field has a richer internal structure built from a tag, two indicators, and one or more subfields.
The tag is the three-digit number that names the field. The two indicators are the first two character positions in the field, and their values are interpreted independently of each other. They supplement the data, for instance by telling the system how many characters to skip when alphabetising a title. An undefined indicator position is left blank.
The subfields break the field into smaller, separately identified pieces of data. Each subfield is preceded by a subfield code, made up of a delimiter (often shown as a dollar sign) followed by a lowercase letter or digit. A practical title field looks like this:
245 10 $a History of Africa : $b from early man to modern times / $c by John Dough & Mary Smith.
Here, 245 is the title tag, and the indicators and subfields organise the parts. The first indicator 1 signals that a title added entry should be generated, while subfield $a holds the main title, $b holds the rest of the title, and $c holds the statement of responsibility. Every field ends with a field terminator, and the very last field of the record closes with both a field terminator and a record terminator, signalling the end of the entire record.
How the three parts work together
Put together, the three components form an efficient retrieval system. The Leader tells the software what the record is and where the data begins. The Directory provides a precise map of every field’s tag, length, and location. The data content delivers the descriptive information through control fields and structured data fields. A system reads them in sequence, and the result is the clean catalogue display you see on screen.
This structure is not just theoretical for libraries here. Open-source library systems like Koha, widely adopted across academic and public libraries, store and exchange their records in MARC 21, which means cataloguers working in these systems are constantly handling Leaders, Directories, and tagged fields, even when the software hides much of the raw structure. Knowing what sits beneath the interface makes troubleshooting and quality control far easier when a record imports incorrectly or a field refuses to display.
What do you think? If the Directory already maps every field’s location, why do you think MARC 21 still keeps key processing details separated out in the fixed-length Leader? And as libraries increasingly move toward linked-data models like BIBFRAME, do you think the rigid three-part MARC structure will remain useful, or will it eventually be replaced?
References
- https://www.loc.gov/marc/specifications/specrecstruc.html
- http://bibliotecas.unr.edu.ar/catalis_pack/catalis/doc/loc/marc/concise/concise.html
- https://help.oclc.org/Metadata_Services/OCLC_MARC_records/Record_structure/020Leader
- https://loc.gov/marc/bibliographic/concise/bdleader.html
- https://www.itsmarc.com/crs/mergedprojects/marcspec/marcspec/leader_record_structure.htm
- https://loc.gov/marc/bibliographic/lite/lbddirectory.html
- https://www.itsmarc.com/crs/mergedProjects/variable_fields_record_structure.htm
- https://www.itsmarc.com/crs/mergedProjects/geninfo/geninfo/content_designators.htm
- https://www.loc.gov/marc/marc21-rda-vendors.html

Leave a Reply