Appendix A — Where can I get MARC records?

Here is a list of data sources I am aware of so far:

A.1 United States of America

• Library of Congress — https://www.loc.gov/cds/products/marcDist.php. MARC21 (UTF-8 and MARC8 encoding), MARCXML formats, open access. Alternative access point: https://www.loc.gov/collections/selected-datasets/?fa=contributor:library+of+congress.+cataloging+distribution+service.
• Harvard University Library — https://library.harvard.edu/open-metadata. MARC21 format, CC0. Institution specific features are documented here
• Columbia University Library — https://library.columbia.edu/bts/clio-data.html. 10M records, MARC21 and MARCXML format, CC0.
• University of Michigan Library — https://www.lib.umich.edu/open-access-bibliographic-records. 1,3M records, MARC21 and MARCXML formats, CC0.
• University of Pennsylvania Libraries — https://www.library.upenn.edu/collections/digital-projects/open-data-penn-libraries. Two datasets are available:
  1. Catalog records created by Penn Libraries 572K records, MARCXML format, CC0,
  2. Catalog records derived from other sources, 6.5M records, MARCXML format, Open Data Commons ODC-BY, use in accordance with the OCLC community norms.
• Yale University — https://guides.library.yale.edu/c.php?g=923429. Three datasets are available:
  1. Yale-originated records: 1.47M records, MARC21 format, CC0
  2. WorldCat-derived records: 6.16M records, MARC21 format, ODC-BY
  3. Other records (MARC21), independent of Yale and WorldCat, where sharing is permitted. 404K records, MARC21 format.
• National Library of Medicine (NLM) catalogue records — https://www.nlm.nih.gov/databases/download/catalog.html. 4.2 million records, NLMXML, MARCXML and MARC21 formats. NLM Terms and Conditions

A.2 Germany

• Deutsche Nationalbibliothek — https://www.dnb.de/DE/Professionell/Metadatendienste/Datenbezug/Gesamtabzuege/gesamtabzuege_node.html (note: the records are provided in utf-8 decomposed). 23.9M records, MARC21 and MARCXML format, CC0.
• Bibliotheksverbundes Bayern — https://www.bib-bvb.de/web/b3kat/open-data. 27M records, MARCXML format, CC0.
• Leibniz-Informationszentrum Technik und Naturwissenschaften Universitätsbibliothek (TIB) — https://www.tib.eu/de/forschung-entwicklung/entwicklung/open-data. (no download link, use OAI-PMH instead) Dublin Core, MARC21, MARCXML, CC0.
• K10plus-Verbunddatenbank (K10plus union catalogue of Bibliotheksservice-Zentrum Baden Würtemberg (BSZ) and Gemensamer Bibliotheksverbund (GBV)) — https://swblod.bsz-bw.de/od/. 87M records, MARCXML format, CC0.

A.3 Others

• Universiteitsbibliotheek Gent — https://lib.ugent.be/info/exports. Weekly data dump in Aleph Sequential format. It contains some Aleph fields above the standard MARC21 fields. ODC ODbL.
• Toronto Public Library — https://opendata.tplcs.ca/. 2.5 million MARC21 records, Open Data Policy
• Répertoire International des Sources Musicales — https://opac.rism.info/index.php?id=8&id=8&L=1. 800K records, MARCXML, RDF/XML, CC-BY.
• ETH-Bibliothek (Swiss Federal Institute of Technology in Zurich) — http://www.library.ethz.ch/ms/Open-Data-an-der-ETH-Bibliothek/Downloads. 2.5M records, MARCXML format.
• British library — http://www.bl.uk/bibliographic/datafree.html#m21z3950 (no download link, use z39.50 instead after asking for permission). MARC21, usage will be strictly for non-commercial purposes.
• Talis — https://archive.org/details/talis_openlibrary_contribution. 5.5 million MARC21 records contributed by Talis to Open Library under the ODC PDDL.
• Oxford Medicine Online (the catalogue of medicine books published by Oxford University Press) — https://oxfordmedicine.com/page/67/. 1790 MARC21 records.
• Fennica — the Finnish National Bibliography provided by the Finnish National Library — http://data.nationallibrary.fi/download/. 1 million records, MARCXML, CC0.
• Biblioteka Narodawa (Polish National Library) — https://data.bn.org.pl/databases. 6.5 million MARC21 records.
• Magyar Nemzeti Múzeum (Hungarian National Library) — https://mnm.hu/hu/kozponti-konyvtar/nyilt-bibliografiai-adatok, 67K records, MARC21, HUNMARC, BIBFRAME, CC0
• University of Amsterdam Library — https://uba.uva.nl/en/support/open-data/data-sets-and-publication-channels/data-sets-and-publication-channels.html, 2.7 million records, MARCXML, PDDL/ODC-BY. Note: the record for books are not downloadable, only other document types. One should request them via the website.
• Portugal National Library — https://opendata.bnportugal.gov.pt/downloads.htm. 1.13 million UNIMARC records in MARCXML, RDF XML, JSON, TURTLE and CSV formats. CC0
• National Library of Latvia National bibliography (2017–2020) — https://dati.lnb.lv/. 11K MARCXML records.

Thanks, Johann Rolschewski, Phú, and Hugh Paterson III for their help in collecting this list! Do you know some more data sources? Please let me know.

There are two more datasource worth mention, however they do not provide MARC records, but derivatives:

• Linked Open British National Bibliography 3.2M book records in N-Triplets and RDF/XML format, CC0 license
• Linked data of Bibliothèque nationale de France. N3, NT and RDF/XML formats, Licence Ouverte/Open Licence