The xd Crossword Corpus

Home of the corpus: https://centuryarcade.org/xdcorpus/

The xd Crossword Corpus is the world’s largest collection of crossword puzzles, with more than 160,000 published crosswords from over 160 publications (as of November 2026), from 1913 to today, kept in a plain-text format for research and preservation. Every puzzle is stored as one .xd file holding its grid, clues, answers and publication details, together with a record of where it came from.

The corpus is maintained by Saul Pwanson and Alex DeJarnatt as a volunteer project of Century Arcade.

Explore

• The Cross(word) Reference: browse and search puzzles, clues and answers across the corpus.
• xd.saul.pw: statistics, reports and corpus-wide analyses, rebuilt nightly.

Use the corpus

Most of the puzzles in the corpus are still under copyright by their publishers and constructors, so the raw corpus itself is not public. It is available for research, in forms ranging from aggregate statistics to derived datasets to the full collection. See ACCESS.md for what data is available, how to ask, and the terms of use.

Contribute puzzles

We welcome puzzles from any publication that meets the collection scope in COLLECTION.md: American-style crosswords, made by a person, part of a published series. If you edit, construct or publish crosswords, CONTRIBUTING.md says what we do with puzzles we are given, who can see them, and what to send. Send puzzles, or questions, to xd@saul.pw. Volunteers who want to convert and submit puzzles themselves are welcome too; see the end of that page.

Policies

• CONTRIBUTING.md: what happens to puzzles you give us, and what to send.
• COLLECTION.md: what the corpus collects.
• EDITORIAL.md: what a corpus puzzle contains and how it is represented.
• ACCESS.md: who may use the corpus, how, and on what terms.
• PRESERVATION.md: how the corpus is kept, and what we commit to.
• GOVERNANCE.md: who maintains the corpus and how decisions are made.
• DECISIONS.md: a dated log of decisions.
• assessments/: yearly self-assessments of the project’s preservation capability.

The parts

xdformat	The .xd file format specification, and converters from .puz and .ipuz.
xd	The pipeline that imports, checks and analyzes the corpus, and builds xd.saul.pw.
The corpus	The corpus itself: a private repository, with the provenance of every puzzle. Access on request.
The Cross(word) Reference	Browsing and search.
Every Puzzle Project	A sister project digitizing historical crosswords from scans, which feeds the corpus.

Citing

If you publish research or results that use the corpus, please cite it:

Pwanson, Saul and DeJarnatt, Alex. (2026). The xd Crossword Corpus. Century Arcade. https://centuryarcade.org/xdcorpus/

Use the year the corpus was accessed. Earlier work cited it as “Pwanson, Saul. (2019). xd Crossword Corpus. http://xd.saul.pw”; that is the same corpus.

Contact

Open an issue in this repository, or find us on Discord.

Acknowledgments

We owe a debt of gratitude to:

• Barry Haldiman, for his diligent collection of thousands of puzzles from the 2000s and 2010s
• David Steinberg and the Pre-Shortzian Puzzle Project (PSPP), for inspiration
• Parker Higgins, for developing and maintaining xword-dl, which has been indispensable for collection
• The hundreds of editors and thousands of constructors who delight us with their creations.
• Arthur Wynne, for kicking the whole thing off on December 21, 1913.