Cataloguing: AACR2, RDA and MARC 21

In short

Cataloguing is the process of describing a library resource and providing the access points (names, titles, subjects) by which a user can find it, so that a catalogue becomes a searchable surrogate of the collection. It has three layers: the descriptive standard (ISBD, AACR2, RDA) that says *what* to record, the conceptual model (FRBR, now IFLA LRM) that says *how bibliographic reality is structured*, and the encoding standard (MARC 21, MARCXML, BIBFRAME) that says *how a machine stores and exchanges it*. RDA replaced AACR2 as the content standard from 31 March 2013 at the Library of Congress, while MARC 21 remains the dominant carrier in Indian ILS software (Koha, SOUL 2.0, LibSys, e-Granthalaya). The topic is heavily examined in UGC-NET/JRF Library and Information Science, university B.LISc/M.LISc papers, and in recruitment tests for librarian posts (KVS, NVS, DSSSB, university and IIT/IIM libraries, CSIR labs, DELNET and INFLIBNET).

Explained

Purpose, objectives and the standards stack

• Catalogue: a list of documents in a library or group of libraries, arranged in a definite order, with an entry giving description plus location (call number).
• Cutter's objects of a catalogue (1876): (1) *finding* — to enable a person to find a book of which the author, title or subject is known; (2) *collocating* — to show what the library has by a given author, on a given subject, in a given kind of literature; (3) *choice* — to assist in the choice of a book by edition or character.
• Paris Principles (ICCP, 1961) standardised choice and form of headings internationally; IFLA Statement of International Cataloguing Principles (ICP, 2009, revised 2016) replaced them and restated the user tasks: find, identify, select, obtain, explore.
• Ranganathan's contribution: *Classified Catalogue Code* (CCC, 5th ed. 1964) with its canons (ascertainability, prepotence, sought heading, context, permanence, currency, etc.), and the *Five Laws of Library Science* as the justification for user-centred cataloguing.
• Standards stack: conceptual model (FRBR 1998 / FRAD / FRSAD → IFLA LRM 2017) → content standard (AACR2 → RDA) → display/punctuation standard (ISBD, consolidated edition 2011) → encoding/exchange (MARC 21, ISO 2709, MARCXML, Dublin Core, BIBFRAME) → vocabularies (LCSH, Sears List, MeSH, DDC/CC/LCC class numbers).
• Types of cataloguing work: original cataloguing, copy (derived) cataloguing by downloading a record via Z39.50/SRU from LC, OCLC WorldCat or IndCat; shared/cooperative cataloguing; centralised cataloguing; CIP (Cataloguing-in-Publication) data printed in the book.

AACR2 — Anglo-American Cataloguing Rules, 2nd edition

• Lineage: Cutter's *Rules for a Printed Dictionary Catalogue* (1876) → AA Code 1908 → ALA Rules 1949 and LC *Rules for Descriptive Cataloging* 1949 → AACR1 (1967), published in separate British and North American texts → AACR2 (1978), a single text built on ISBD → revisions 1988, 1998, 2002 revision with updates to 2005 (the final text). Published jointly by ALA, CILIP (LA) and the Canadian Library Association; governed by the Joint Steering Committee (JSC, later RSC).
• Core principle: *describe the item in hand*; description is based on the chief source of information (for a book, the title page), with prescribed sources for each area.
• Structure — Part I: Description (Ch. 1 general rules; 2 books, pamphlets and printed sheets; 3 cartographic materials; 4 manuscripts; 5 music; 6 sound recordings; 7 motion pictures and videorecordings; 8 graphic materials; 9 electronic resources; 10 three-dimensional artefacts and realia; 11 microforms; 12 continuing resources; 13 analysis).
• Structure — Part II: Headings, uniform titles and references (Ch. 21 choice of access points; 22 headings for persons; 23 geographic names; 24 headings for corporate bodies; 25 uniform titles; 26 references). Appendices: A capitalisation, B abbreviations, C numerals, D glossary.
• The eight ISBD areas of description: 1 Title and statement of responsibility; 2 Edition; 3 Material (or type of publication) specific details; 4 Publication, distribution, etc.; 5 Physical description; 6 Series; 7 Notes; 8 Standard number and terms of availability. (ISBD consolidated 2011 added Area 0, content form and media type.)
• Three levels of description (rule 1.0D): first level = minimal (title proper / first statement of responsibility, edition, publisher, date, extent, notes, ISBN); second level = normal library practice; third level = all applicable elements.
• Main entry and added entries: the main entry is the complete record under the chosen primary access point; added entries give secondary access. Choice (Ch. 21): single personal author → that person; shared responsibility of two or three → the first named, with added entries; more than three with no principal author → title main entry plus an added entry for the first named (the "rule of three"); works of unknown or diffuse authorship, collections with a collective title, and works by editors/compilers → title; corporate body only in the specific categories of 21.1B2 (administrative works on the body itself, certain legal and official works, collective thought of the body, reports of collective activity, performing-group works, cartographic works of the body).
• GMD (general material designation) such as *[sound recording]*, *[electronic resource]*, *[videorecording]* was given after the title proper.
• Latin and other abbreviations were prescribed: *[S.l.]* (sine loco, place unknown), *[s.n.]* (sine nomine, publisher unknown), *et al.*, *ill.*, *p.*, *2nd ed.*, *c1998*, *[n.d.]*.
• Strengths: mature, internationally tested, compact records, strong authority practice. Limitations: class-of-material (format-first) structure that struggled with digital and multi-version resources; abbreviation and the rule of three suppressed data the user wanted; no explicit conceptual model; main-entry thinking is a card-catalogue legacy; English-language/Anglo-American bias.

RDA — Resource Description and Access

• What it is: the content standard that succeeded AACR2, released in the web-based RDA Toolkit in June 2010; implemented by LC, the British Library, Library and Archives Canada and the National Library of Australia on 31 March 2013. It is a *content* standard, not an encoding standard and not a display standard — it can be carried in MARC 21, MARCXML, BIBFRAME or linked-data triples.
• Model basis: originally FRBR + FRAD, with the WEMI entities — Work (the intellectual creation), Expression (its realisation in a particular form/language), Manifestation (its physical or digital embodiment), Item (a single exemplar) — plus Person, Family, Corporate Body, Concept, Object, Event, Place. After the 3R Project (RDA Toolkit Restructure and Redesign), the new Toolkit text became the official RDA on 15 December 2020 and is aligned to IFLA LRM (2017), which adds *Agent, Collective Agent, Nomen, Place, Timespan* and the concept of a representative expression.
• Structure: 10 sections and 37 chapters — Sections 1–4 record *attributes* (manifestation/item, work/expression, agents, concepts etc.); Sections 5–10 record *relationships*; appendices cover capitalisation, abbreviations, initial articles and the relationship designators (author, editor, illustrator, translator, container of, contained in …).
• Key RDA principles:
  • "Take what you see, accept what you get" — transcribe from the source without correcting or abbreviating; errors are transcribed and explained in a note (AACR2's *[sic]*/bracketed correction practice is dropped for most elements).
  • No prescribed abbreviations — *Second edition*, *pages*, *illustrations*, *[Place of publication not identified]*, *[publisher not identified]*.
  • Rule of three abolished — record all statements of responsibility; only the first is core, the rest optional.
  • GMD replaced by three elements: 336 content type (text, still image, performed music, cartographic image…), 337 media type (unmediated, audio, computer, video, microform), 338 carrier type (volume, audio disc, online resource, videodisc).
  • Core elements: title proper, statement of responsibility (first), edition statement, publication statement (place, publisher, date), series, identifier, content type and carrier type, extent; LC-PCC practice also records media type.
  • Relationship designators and authorised access points replace rigid main-entry logic; "preferred title/name" replaces "uniform title/heading"; "creator" replaces "author" as the general term.
• Advantages: content/carrier separated, so FRBR-style clustering of all versions of a work is possible; format-neutral and extensible to born-digital resources; fuller transcription aids identification; designed for linked data (RDA Registry publishes the element set and vocabularies as URIs).
• Limitations: training and retrospective-conversion cost; MARC 21 can carry RDA data but cannot exploit the relationships; the Toolkit is a paid subscription; hybrid AACR2/RDA catalogues cause inconsistent displays; instructions are often options rather than rules, so local policy statements (LC-PCC PS) are needed.

MARC 21 — the encoding and exchange standard

• Origin: the MARC (MAchine-Readable Cataloging) pilot was led by Henriette Avram at the Library of Congress in 1966–68; MARC II became the basis of national formats; MARC 21 (1999) harmonised USMARC and CAN/MARC and is maintained by LC with Library and Archives Canada.
• Five formats: Bibliographic, Authority, Holdings, Classification, Community Information.
• Record structure (ISO 2709 / ANSI-NISO Z39.2): three parts —
  1. Leader — 24 fixed character positions (record length 00-04, record status 05, type of record 06, bibliographic level 07, encoding level 17, descriptive cataloguing form 18 = i for ISBD-punctuated RDA records, a for AACR2, base address of data 12-16).
  2. Directory — one 12-character entry per field: 3-character tag + 4-digit field length + 5-digit starting character position.
  3. Variable fields — control fields 001–009 (no indicators or subfields) and data fields 010–999, each with two indicators (values 0–9 or blank) and subfields introduced by a delimiter ($, ‡ or |) plus a one-character subfield code.
• Tag blocks: 0XX control numbers and codes; 1XX main entry; 2XX title, edition, imprint; 3XX physical description and RDA content/media/carrier; 4XX series statements; 5XX notes; 6XX subject access; 7XX added and linking entries; 8XX series added entries and electronic location; 9XX locally defined.
• Frequently examined bibliographic fields: 001 control number; 005 date and time of latest transaction; 008 fixed-length data elements (40 positions: date entered, publication date, place, language, illustrations, literary form); 020 ISBN; 022 ISSN; 040 cataloguing source ($e rda); 041 language code; 082 DDC number; 100/110/111 main entry personal/corporate/meeting name; 130 main entry uniform title; 245 title statement (indicator 1 = added entry for title, indicator 2 = number of non-filing characters for the initial article); 250 edition; 260 publication (AACR2) / 264 production, publication, distribution, manufacture, copyright (RDA, with indicator 2: 0 production, 1 publication, 2 distribution, 3 manufacture, 4 copyright); 300 physical description; 336/337/338 content, media, carrier type; 490 series statement; 5XX notes (500 general, 504 bibliography, 505 contents, 520 summary); 600/610/650/651 subject added entries (indicator 2 = thesaurus: 0 LCSH, 2 MeSH, 7 source in $2); 700/710 added entries with $e relationship designator; 856 electronic location and access (URL in $u).
• Authority format fields: 1XX authorised heading, 4XX *see from* tracings (variants), 5XX *see also from* tracings (related headings), 667/670/675 cataloguer's notes and sources consulted. Authority control keeps one authorised form per entity, with references from variants — the basis of reliable collocation in the OPAC.
• Serialisations and successors: MARCXML (XML wrapper for MARC 21), MODS and MADS (richer XML), Dublin Core (15 elements, used in DSpace repositories such as Shodhganga and NDLI), BIBFRAME 2.0 (LC's RDF linked-data replacement for MARC, with Work, Instance, Item) and Schema.org for discoverability.
• Limitations of MARC 21: flat, record-based and string-based (data as text rather than identifiers), redundancy across records, 1960s fixed-field economy, no native way to express WEMI relationships, invisible to the general web — the reasons BIBFRAME exists. Its strengths are universal ILS support, decades of shared records and robust exchange.

Indian standards, practice and cataloguing codes

• Codes used in India: CCC (Ranganathan) is still taught and examined and used in some Indian libraries; AACR2 is the historic mainstream; RDA adoption is growing in national institutions, IITs/IIMs and large university libraries, with many catalogues still hybrid AACR2/RDA.
• CCF (Common Communication Format) — UNESCO/PGI exchange format, used by Indian agencies and in older INFLIBNET/DESIDOC practice; MARC 21 is now the normal format in Indian ILS software.
• Software: SOUL 2.0 (INFLIBNET, MARC 21 compliant), Koha (open source, MARC 21 or UNIMARC, Z39.50/SRU import), e-Granthalaya (NIC), LibSys/LSEase, NewGenLib; repositories use DSpace/EPrints with Dublin Core.
• Union catalogues and networks: IndCat (INFLIBNET's union catalogue of books, theses and serials), DELNET union catalogues, Shodhganga (ETDs, mandated metadata under the UGC regulations on plagiarism/thesis deposit), National Digital Library of India, NLK (National Library, Kolkata) catalogue.
• Subject vocabularies: Sears List of Subject Headings (school and small libraries), LCSH (academic), MeSH (medical, e.g. AIIMS/NIC), and chain indexing (Ranganathan's method) for classified catalogues.
• Interoperability protocols: Z39.50, SRU/SRW, OAI-PMH for metadata harvesting, ISO 2709 for record exchange. For Indian-language cataloguing, Unicode (ISCII superseded in practice) allows original-script 880 fields with vernacular transliteration.

Key points

• Catalogue – an ordered list of a library's documents, each entry giving description plus location.
• Cutter's objects – finding, collocating, choice; restated by ICP 2016 as find, identify, select, obtain, explore.
• ISBD areas – eight areas (plus Area 0 since 2011) with prescribed punctuation, giving language-independent display.
• AACR2 – 1978, final text the 2002 revision with 2005 updates; Part I description, Part II headings; class-of-material arrangement.
• Chief source of information – the title page for a book; description must come from prescribed sources.
• Main entry – the primary access point under which the full record is made; added entries give further access.
• Rule of three – AACR2 limited statements of responsibility to three names; RDA abolished it.
• FRBR/WEMI – Work, Expression, Manifestation, Item; the model behind RDA; extended by IFLA LRM (2017).
• RDA – released 2010, implemented 31 March 2013; the 3R-redesigned, LRM-aligned text became official on 15 December 2020.
• Transcription rule – "take what you see": no prescribed abbreviations, errors transcribed and noted.
• GMD replaced – by content type (336), media type (337) and carrier type (338).
• MARC 21 – 1999 harmonisation of USMARC and CAN/MARC; five formats; maintained by LC.
• MARC record structure – 24-character Leader, 12-byte-per-field Directory, variable fields with 2 indicators and $ subfields; ISO 2709.
• 245 – title statement; indicator 2 counts non-filing characters of the initial article ("The " = 4).
• 260 vs 264 – AACR2 imprint vs RDA publication statement with indicator 2 for function.
• Authority control – one authorised access point per entity, 4XX see-from and 5XX see-also-from references.
• BIBFRAME 2.0 – LC's RDF linked-data model (Work, Instance, Item) intended to succeed MARC.
• Indian practice – MARC 21 in Koha/SOUL 2.0/e-Granthalaya, Dublin Core in DSpace, CCF historically, CCC still taught.

Examples

1. The same book described under AACR2 and under RDA. Title page reads: *Introduction to Knowledge Organisation*, Second Edition, by Amitabha Basu, Sunita Rao, K. Menon and R. Iyer; Oxford University Press, New Delhi, 2024; xii, 318 pages, illustrated, 24 cm; ISBN 978-81-265-1123-4.

2. The MARC 21 record for the same book.

=LDR  00000nam a2200000 i 4500
=008  240115s2024    ii a     b    000 0 eng d
=020  \\$a9788126511234
=040  \\$aIN-ABL$beng$erda$cIN-ABL
=082  04$a025.4$223
=100  1\$aBasu, Amitabha,$eauthor.
=245  10$aIntroduction to knowledge organisation /$cAmitabha Basu, Sunita Rao, K. Menon and R. Iyer.
=250  \\$aSecond edition.
=264  \1$aNew Delhi :$bOxford University Press,$c2024.
=300  \\$axii, 318 pages :$billustrations ;$c24 cm
=336  \\$atext$btxt$2rdacontent
=337  \\$aunmediated$bn$2rdamedia
=338  \\$avolume$bnc$2rdacarrier
=504  \\$aIncludes bibliographical references and index.
=650  \0$aInformation organization.
=700  1\$aRao, Sunita,$eauthor.
=856  40$uhttps://opac.example.ac.in/record/12345

Leader/06 = a (language material), /07 = m (monograph), /18 = i (ISBD punctuation, RDA); 040 $e rda declares the content standard; 100 indicator 1 = 1 (surname entry); 245 indicator 1 = 1 (title added entry made), indicator 2 = 0 (no initial article to skip).

3. Computing the base address of data. A record has the 24-byte Leader and 14 variable fields. Directory = 14 × 12 = 168 bytes, plus 1 byte field terminator = 169. Base address of data (Leader/12-16) = 24 + 169 = 193, written as 00193. If the variable-field area totals 842 bytes and the record terminator is 1 byte, total record length = 193 + 842 + 1 = 1036, written in Leader/00-04 as 01036.

4. Non-filing indicator. For 245 14$aThe economics of Indian libraries, indicator 2 = 4 — three letters of "The" plus one space — so the OPAC files the title under "E". For a Hindi title beginning "Ek ", the count is 3. A wrong count is the single commonest MARC filing error.

5. Choice of main entry under AACR2 versus RDA. A conference proceedings volume: *Proceedings of the National Seminar on Digital Preservation, held at University of Delhi, 2024*, edited by P. Krishnan.

• AACR2: a named conference is a corporate body (21.1B2) → main entry under National Seminar on Digital Preservation (2024 : Delhi, India) in field 111; added entry 700 for Krishnan as editor.
• RDA: the conference is the creator, so the authorised access point for the work is the same corporate body, but Krishnan is linked with the relationship designator $e editor, and no "main entry" concept is invoked — the record simply carries one creator and several related agents.
• If instead there were no named conference, the work would be entered under title, with 700 $e editor for Krishnan under both codes.

Where it is used

• Academic and research libraries: cataloguing sections of central libraries in IITs, IIMs, IISc, central and state universities; original and copy cataloguing in Koha or SOUL 2.0, with records contributed to IndCat.
• National and special libraries: National Library Kolkata, DESIDOC, NISCPR, ICAR and ICMR institute libraries, Parliament Library, CSIR labs — MARC 21 bibliographic and authority records, MeSH/LCSH subject work.
• Networks and consortia: INFLIBNET (SOUL development, union catalogues, Shodhganga ETD metadata), DELNET inter-library loan built on union catalogue records, e-ShodhSindhu.
• Library software and metadata industry: Koha/NewGenLib implementers, LibSys, Informatics India, Nielsen/ProQuest/EBSCO record supply, ONIX-to-MARC conversion for publishers.
• Job roles and recruitment: Library Assistant, Semi-Professional Assistant, Professional Assistant, Assistant/Deputy Librarian — UGC-NET LIS Paper II has a whole unit on knowledge organisation and metadata standards; KVS/NVS/DSSSB/RRB librarian papers ask MARC tags and AACR2/RDA differences directly.
• Digital libraries and repositories: DSpace institutional repositories and NDLI use Dublin Core and OAI-PMH, with crosswalks from MARC 21; linked-data and BIBFRAME pilots in large libraries.

Interview questions

• Q: What is the difference between a content standard and an encoding standard? A: RDA (and earlier AACR2) tells the cataloguer what data to record and how to word it; MARC 21 tells the system how to store and exchange that data. A record can be RDA in content and MARC 21 in encoding.
• Q: Name the four FRBR Group 1 entities with an example. A: Work (Ramayana as an intellectual creation), Expression (an English verse translation), Manifestation (the 2024 OUP edition of that translation), Item (the copy with accession number 4521).
• Q: What replaced the GMD in RDA, and why? A: Three separate elements — content type (336), media type (337) and carrier type (338) — because a single phrase like "electronic resource" conflated what the content is, what device is needed and what the carrier is.
• Q: What is the rule of three and what happened to it? A: AACR2 recorded at most three names in a statement of responsibility, using "… [et al.]" beyond that; RDA abolished the limit — all can be recorded, with only the first as a core element.
• Q: Explain the three parts of a MARC 21 record. A: A 24-character Leader of coded values, a Directory with a 12-character entry (tag, field length, starting position) for each field, and the variable fields themselves with indicators and subfields, terminated per ISO 2709.
• Q: What does the second indicator of field 245 mean? A: The number of non-filing characters at the start of the title — the initial article plus the following space — so the catalogue files under the first significant word.
• Q: Why was 264 introduced when 260 existed? A: RDA treats production, publication, distribution, manufacture and copyright as distinct statements; 264's second indicator identifies which function is being recorded, which 260 could not do.
• Q: What is authority control and why does it matter? A: Maintaining one authorised access point per person, body, place or work, with references from variant forms, so that all works by an entity collocate despite spelling, language or name changes.
• Q: Is RDA tied to MARC? A: No. RDA is format-neutral; its data can be encoded in MARC 21, MARCXML, BIBFRAME or RDF, which is why the RDA Registry publishes its elements and vocabularies as URIs.
• Q: What is BIBFRAME and how does it differ from MARC? A: An RDF linked-data model from the Library of Congress with Work, Instance and Item; it uses URIs and relationships instead of text strings in a flat record, making library data usable on the web.
• Q: Which cataloguing code did Ranganathan author, and how does its approach differ? A: The Classified Catalogue Code, designed for a classified catalogue with chain indexing and governed by canons, whereas AACR2 was designed for a dictionary/alphabetical catalogue.
• Q: A book's title page misspells a word. How do AACR2 and RDA handle it? A: AACR2 would transcribe with *[sic]* or a bracketed correction; RDA transcribes as found and explains the error in a note, optionally adding a variant title access point.

Common mistakes

• Calling RDA "a new MARC" or "a classification scheme" — it is a content standard, independent of both encoding and classification.
• Writing a 245 second indicator of 3 for "The " (it is 4 — three letters plus the space), or counting an article in a language where it is a filing word.
• Mixing AACR2 and RDA conventions in one record — "2nd ed." with 336/337/338, or $e rda in 040 alongside *[S.l.] : [s.n.]*.
• Omitting the subfield delimiter or the $2 source code in 336/337/338, and forgetting that 650 second indicator must declare the thesaurus.
• Confusing main entry (a complete record under the primary access point) with added entry, or assuming RDA still requires a main entry.
• Quoting FRBR as current when IFLA LRM (2017) has consolidated FRBR, FRAD and FRSAD, and the LRM-aligned RDA text has been official since 15 December 2020.
• Treating the chief source of information as "anywhere in the book" — description must come from the prescribed source, with data from elsewhere bracketed.

Related topics

Library classification: DDC and Colon Classification · Information retrieval and indexing · Metadata standards, Dublin Core and digital repositories · Library automation and integrated library systems · Union catalogues, resource sharing and library networks in India · Subject headings and thesaurus construction