Skip to content

MARC Primer

An introduction to MARC record structure for developers.

What is MARC?

MARC (MAchine-Readable Cataloging) is a data format for bibliographic records. Created by the Library of Congress in the 1960s, it remains the standard for library systems worldwide.

A MARC record describes a single item—a book, journal, map, recording, or other material—with structured metadata about its title, author, publisher, subjects, and more.

Why MARC matters:

  • Over 200 million records in WorldCat alone
  • Standard interchange format for library systems (ILS, OCLC, Z39.50)
  • Most library metadata exists in MARC format

Record Structure

A MARC record has three parts:

┌─────────────────────────────────────────┐
│  Leader (24 bytes)                      │
├─────────────────────────────────────────┤
│  Directory (variable length)            │
├─────────────────────────────────────────┤
│  Fields (variable length)               │
└─────────────────────────────────────────┘

Leader

The leader is exactly 24 characters containing record metadata:

Position Length Description
00-04 5 Record length
05 1 Record status (n=new, c=corrected, d=deleted)
06 1 Type of record (a=language material, e=map, etc.)
07 1 Bibliographic level (m=monograph, s=serial)
08-09 2 Type of control
10 1 Indicator count (always "2")
11 1 Subfield code count (always "2")
12-16 5 Base address of data
17-19 3 Encoding level, descriptive form, linked record
20-23 4 Entry map (always "4500")

Python: 

leader = record.leader
print(leader.record_status)  # 'n', 'c', or 'd'
print(leader.type_of_record)  # 'a' for language material

Rust: 

let leader = record.leader();
println!("{}", leader.record_status());
println!("{}", leader.type_of_record());

Directory

The directory is an index listing each field's tag, length, and starting position. MRRC handles directory parsing automatically—you work directly with fields.

Field Types

MARC has two types of fields:

Control Fields (001-009)

Control fields contain unstructured data—no indicators or subfields.

Tag Name Example
001 Control Number ocm12345678
003 Control Number Identifier OCoLC
005 Date/Time Last Modified 20240115120000.0
006 Fixed-Length Data Elements (additional material)
007 Physical Description (format-specific codes)
008 Fixed-Length Data Elements (40 character coded data)

Python: 

control_number = record["001"]
if control_number:
    print(control_number.value)  # "ocm12345678"

Rust: 

if let Some(field) = record.field("001") {
    if let Some(data) = field.data() {
        println!("{}", data);
    }
}

Data Fields (010-999)

Data fields have a 3-digit tag, two indicators, and one or more subfields.

Tag   Ind1  Ind2  Subfields
245   1     0     $a Title : $b subtitle / $c author.

Indicators

Each data field has two single-character indicators that modify the field's meaning:

Field Ind1 Ind2 Meaning
245 (Title) 0-1 0-9 Title added entry; Nonfiling characters
100 (Author) 0-3 (blank) Type of name
650 (Subject) (blank) 0-7 Level; Thesaurus

Example: 245 14 means: - Ind1=1: Title added entry - Ind2=4: Skip first 4 characters when filing ("The " in "The Book")

Subfields

Subfields are identified by a delimiter ($ or ‡) and a single-character code:

245 10 $a Programming in Rust :$b a comprehensive guide /$c by Jane Smith.
       ↑                        ↑                         ↑
       $a = title proper        $b = subtitle             $c = statement of responsibility

Python: 

title_field = record["245"]
if title_field:
    print(title_field["a"])  # "Programming in Rust :"
    print(title_field["b"])  # "a comprehensive guide /"
    print(title_field["c"])  # "by Jane Smith."
    print(title_field.ind1)  # "1"
    print(title_field.ind2)  # "0"

Rust: 

if let Some(field) = record.field("245") {
    println!("{:?}", field.subfield("a"));
    println!("{:?}", field.subfield("b"));
    println!("Indicators: {} {}", field.indicator1(), field.indicator2());
}

Common Fields

Identification

Tag Name Subfields
001 Control Number (control field, no subfields)
010 LC Control Number $a number
020 ISBN $a ISBN, $q qualifier
022 ISSN $a ISSN
035 System Control Number $a number

Main Entry (Author)

Tag Name Subfields
100 Personal Name $a name, $d dates, $e relator
110 Corporate Name $a name, $b subordinate unit
111 Meeting Name $a name, $d date, $c location

Title

Tag Name Subfields
245 Title Statement $a title, $b subtitle, $c responsibility
246 Varying Form of Title $a title, $b remainder
130/240 Uniform Title $a title, $l language

Publication

Tag Name Subfields
260 Publication (older) $a place, $b publisher, $c date
264 Production/Publication $a place, $b name, $c date

Physical Description

Tag Name Subfields
300 Physical Description $a extent, $b details, $c dimensions

Subject Access

Tag Name Subfields
600 Subject - Personal $a name, $d dates, $t title
610 Subject - Corporate $a name, $b subordinate
650 Subject - Topical $a topic, $x general subdivision
651 Subject - Geographic $a place, $x subdivision

Added Entries

Tag Name Subfields
700 Added Entry - Personal $a name, $e relator
710 Added Entry - Corporate $a name, $b subordinate
856 Electronic Location $u URI, $z note

Using Convenience Methods

MRRC provides shortcuts for common fields:

Python: 

# Instead of parsing 245$a yourself:
print(record.title())

# Instead of checking 100, 110, 111:
print(record.author())

# Gets all ISBNs from 020$a:
for isbn in record.isbns():
    print(isbn)

Rust: 

println!("{:?}", record.title());
println!("{:?}", record.author());
for isbn in record.isbns() {
    println!("{}", isbn);
}

Character Encoding

MARC records use one of two encodings:

Encoding Leader position 09 Description
MARC-8 (blank) Legacy encoding with escape sequences
UTF-8 a Unicode, modern standard

MRRC handles encoding automatically, converting MARC-8 to UTF-8 when reading.

ISO 2709 Binary Format

MARC records are commonly stored in ISO 2709 format (.mrc files):

  • Records are concatenated with no separator
  • Each record ends with a record terminator (ASCII 29)
  • Fields within a record end with a field terminator (ASCII 30)
  • Subfields are delimited by ASCII 31

MRRC handles all parsing automatically—you work with structured Record objects.

See Also