PIERCE: Parsed Index of Editions, Rules, and Codified Evolution in Sport (Release 2026.15)
Authors/Creators
Description
PIERCE: Parsed Index of Editions, Rules, and Codified Evolution in Sport
Release 2026.15 (version of record)
PIERCE is a research database of official sport rule books, decomposed to the level of the individual rule provision, with frozen public identifiers, provenance to the registered source documents, and full-text search. Live interface: https://adamcopeland.co/rules
This deposit is the complete public export of release 2026.15, cut 3 October 2026. Database content sha256 (first 16): 2db5acd93908d32e (full value in data/manifest.json).
Corpus at this cut (filled from this release's own export, data/totals.json and data/manifest.json, 3 October 2026): 348 editions on record, across 12 sports and 43 series (1892-2026). Of these, 346 are exported (144,535 rule nodes, 14,277,766 words of body text). The other 2 are retired: fa-soccer-2009-10 into fa-soccer-2009 and fa-soccer-2014 into fa-soccer-2013, a merge made on 2026-09-06. The database also holds 607 registered source documents and 26 governing bodies. Counting every rule row it still holds, including the two retired editions', the underlying totals are 148,901 rule rows holding 14,878,841 words, and 2,572 approved-ruling rows. In this README, "words" means the exporter's token count of body text, the figure in data/totals.json. A plain whitespace count of the same text comes out slightly higher.
What changed since 2026.14
This release follows 2026.14 (deposited 16 September 2026). It carries these kinds of change: new rule books, repairs to older books, retired and new permanent identifiers, two changes to how the public reader shows text, a new cross-reference table (which is in the database but not in this deposit), and corrections to two documents of release 2026.14.
1. New rule books
The release adds 32 rule books (edition ids 338 to 369). They add 23,123 rule rows. All 32 new books carry the text policy truncated (first sentence per provision).
By sport:
- Australian Football: 1 book, 260 rule rows. AFL Laws of the Game: 2026 (edition 341).
- Baseball: 2 books, 292 rule rows. MLB Baseball: 2024 (edition 342); NCAA Baseball Rules: 2025-26 (edition 366).
- Basketball: 4 books, 560 rule rows. NBA Official Rules: 2025-26 (edition 351); WNBA Official Rules: 2026 (edition 352); NCAA Basketball: 2026-27 (edition 367); NCAA Basketball (Women): 2025-27 (edition 368).
- Bowling: 5 books, 1,675 rule rows. PBA Tour Rule Book: 2025 (edition 338), 2026 (edition 339); USBC Playing Rules: 2025-26 (edition 354), 2026-27 (edition 356); NCAA Women's Bowling: 2026 (edition 359).
- Football: 1 book, 429 rule rows. NCAA Football: 2026 (edition 369).
- Hockey: 3 books, 1,587 rule rows. NCAA Hockey: 2026-28 (edition 344); NHL Hockey: 2025-26 (edition 347); IIHF Official Rulebook: 2025-26 (edition 353).
- Lacrosse: 4 books, 1,707 rule rows. World Lacrosse Men's Field Lacrosse Rules: 2025-27 (edition 357); World Lacrosse Women's Field Lacrosse Rules: 2025-26 (edition 358); World Lacrosse Box Lacrosse Rules: 2026-28 (edition 360); NCAA Women's Lacrosse Rules: 2026-27 (edition 365).
- Pickleball: 1 book, 577 rule rows. USA Pickleball Rulebook: 2026 (edition 361).
- Soccer: 5 books, 13,083 rule rows. IFAB Soccer: 2024-25 (edition 343); Premier League Handbook: 2023-24 (edition 348), 2024-25 (edition 349), 2025-26 (edition 350); NCAA Soccer Rules: 2026-27 (edition 364).
- Tennis: 3 books, 396 rule rows. ATP Tour Rulebook: 2026 (edition 345); WTA Tour Rulebook: 2026 (edition 346); ITA Collegiate Tennis Rulebook: 2025-26 (edition 355).
- Track & Field: 3 books, 2,557 rule rows. NCAA Track & Field and Cross Country: 2025 (edition 340); World Para Athletics Rules and Regulations: 2026 (edition 362); World Athletics Competition and Technical Rules: 2026 (edition 363).
The new books open 13 new series and bring 8 new governing bodies. The new series are:
- NBA Official Rules (Basketball, National Basketball Association): 1 book
- WNBA Official Rules (Basketball, Women's National Basketball Association): 1 book
- IIHF Official Rulebook (Hockey, International Ice Hockey Federation): 1 book
- USBC Playing Rules (Bowling, United States Bowling Congress): 2 books
- ITA Collegiate Tennis Rulebook (Tennis, Intercollegiate Tennis Association): 1 book
- World Lacrosse Men's Field Lacrosse Rules (Lacrosse, World Lacrosse): 1 book
- World Lacrosse Women's Field Lacrosse Rules (Lacrosse, World Lacrosse): 1 book
- World Lacrosse Box Lacrosse Rules (Lacrosse, World Lacrosse): 1 book
- World Para Athletics Rules and Regulations (Track & Field, World Para Athletics): 1 book
- World Athletics Competition and Technical Rules (Track & Field, World Athletics): 1 book
- NCAA Soccer Rules (Soccer, National Collegiate Athletic Association): 1 book
- NCAA Women's Lacrosse Rules (Lacrosse, National Collegiate Athletic Association): 1 book
- NCAA Baseball Rules (Baseball, National Collegiate Athletic Association): 1 book
The new governing bodies are: National Basketball Association (NBA); Women's National Basketball Association (WNBA); International Ice Hockey Federation (IIHF); United States Bowling Congress (USBC); Intercollegiate Tennis Association (ITA); World Lacrosse (WL); World Para Athletics (WPA); World Athletics (WA).
Of the new series, 3 are National Collegiate Athletic Association (NCAA) series: soccer, women's lacrosse and baseball. Of the new books, 9 come from the NCAA, including new books in series that already existed (basketball, football, women's bowling, hockey, track and field). The full list, with sport, governing body, title, year, edition id, public id and row count, is in changes/new_editions.csv.
Abbreviations in the names of books and series (the governing body that publishes the rules): AFL (Australian Football League), ATP (Association of Tennis Professionals), FA (The Football Association), IFAB (International Football Association Board), IIHF (International Ice Hockey Federation), ITA (Intercollegiate Tennis Association), MLB (Major League Baseball), NBA (National Basketball Association), NCAA (National Collegiate Athletic Association), NFL (National Football League), NHL (National Hockey League), PBA (Professional Bowlers Association), USBC (United States Bowling Congress), WA (World Athletics), WL (World Lacrosse), WNBA (Women's National Basketball Association), WPA (World Para Athletics), WTA (Women's Tennis Association).
Three small errors repaired on 3 October 2026. Before this release was cut, three small errors in the newest books were repaired in the database. All three books are new in this release, and no identifier changed.
- Premier League Handbook 2025-26: the first word of clause 1 of the Commercial Rights Protocol (
premier-league-handbook-2025-26/APP.fmv_commercial_rights.1) was wrong. The word "This" was stored as "his", with a stray "T" at the end of the line. The clause now opens with its right first word. - Premier League Handbooks 2024-25 and 2025-26: the page header (the title of the part, printed at the top of each page) sat inside clause 13 of the Constitution of The Professional Football Negotiating and Consultative Committee (
Misc.PFNCC.13in each book). It is removed. - NHL 2025-26: rule 37.6 (
nhl-hockey-2025-26/37.6) had a line break inside "On- and Off-Ice Officials". It is now a space.
Further repairs in the evening of 3 October 2026. The database was repaired again later that day, before this release was cut. The repairs in books that are new in this release are listed here; those in books that were in 2026.14 are in section 3. The row counts come from comparing the database before these repairs with the database of this release.
- Premier League Handbooks 2023-24, 2024-25 and 2025-26: the second line of the page header (the title of the part, printed at the top of each page) sat inside 9 more clauses of the Constitution of The Professional Football Negotiating and Consultative Committee: clauses 3, 7 and 11 of each book (
Misc.PFNCC.3,Misc.PFNCC.7andMisc.PFNCC.11). It is removed. A stray letter "C." was also removed from clause 1 of the Commercial Rights Protocol (APP.fmv_commercial_rights.1) in the 2024-25 and 2025-26 handbooks (2 rows). In all, 11 rows changed, text only, and no identifier changed. - NHL 2025-26: rule 11.2 (
nhl-hockey-2025-26/11.2) had a line break after the hyphen in "non-compressed". The line break is gone and the hyphen stays (1 row). - NHL 2025-26: the referee signals table of Rule 29. The book prints 35 signals in a table. The database held the table as 4 rows, and most of the table's words sat in one of them (rule 29.3). It now holds one row for each printed signal,
nhl-hockey-2025-26/29.1tonhl-hockey-2025-26/29.35. 4 existing rows were corrected, and 32 rows were added (nhl-hockey-2025-26/29.4tonhl-hockey-2025-26/29.35), each with a new permanent identifier. The stray words of the next section's title were taken out of Rule 29. No identifier was retired.
2. Premier League handbooks and their clause rows
The Premier League Handbook series gained 3 books. Each was loaded once, down to its numbered rules, and then extended in steps. The appendices and the miscellaneous sections were added later, first as one row per item and then as one row per numbered clause (a clause row is one numbered paragraph of an appendix or of a miscellaneous section).
- Premier League Handbook 2023-24 (edition 348): 3,940 rule rows, of which 413 were added after the first load.
- Premier League Handbook 2024-25 (edition 349): 4,108 rule rows, of which 409 were added after the first load.
- Premier League Handbook 2025-26 (edition 350): 4,290 rule rows, of which 446 were added after the first load.
In all, 1,268 rows were added after the first load. Of them, 1,158 are clause rows. The steps, by the day the rows were issued:
- 2026-09-24: 98 rows, of which 0 clause rows
- 2026-09-26: 730 rows, of which 730 clause rows
- 2026-09-29: 130 rows, of which 130 clause rows
- 2026-09-30: 310 rows, of which 298 clause rows
These steps retired no permanent identifier (0 retired in these three books). Every added row is listed, with its permanent identifier, in changes/added_after_first_ingest_rows.csv. The steps are summarised in changes/added_after_first_ingest_batches.csv. Both lists also hold the 32 rows added to NHL 2025-26 on the evening of 3 October 2026 (the referee signals, section 1). They are not Premier League rows and are not counted above.
3. Repairs to books that were in 2026.14
Of the books that were in 2026.14, 34 changed. In them, 510 rule rows changed their body text and 39 rows changed their heading. Also, 116 rows changed place in their book (their parent, their position or their path label). Apart from the two retirements in section 4, no text repair changed a rule-row identifier: a repaired row keeps its identifier. The count of changed rows for every book is in changes/changed_editions.csv.
Worked examples and cross-references. Some books print worked examples (approved rulings). The database stores them as their own records. Four repairs moved text between rule text and those records.
- Cross-references put back. The 12 September 2026 split had cut 46 passages out of rule text that were cross-references, not worked examples. They are back in rule text: 31 in NCAA football 2001 to 2005, 2 in NFL 2008 and 13 in NCAA men's lacrosse. In NCAA football 2001 to 2005 and NFL 2008 (5 books), the number of rule rows that changed is 33.
- NCAA men's lacrosse 2012 to 2024 (7 books). The rule that finds where a worked example ends was corrected, so that a section's own clauses printed after an example stay in rule text. Rule rows changed in these books: 120 (this count also holds the cross-reference rows above and the section 6.5 repair below). Ruling ids retired: 66, because their records merged into the record printed above them. Ruling ids issued: 60.
- NCAA men's lacrosse 2025. This book came into 2026.14 without the split of its worked examples. It is split now. Ruling records created: 63. Rule rows that changed text: 51.
- NCAA women's bowling 2012 to 2024 (7 books). The same corrected end-of-example rule was applied to bowling. Rule rows changed: 86. Ruling ids retired: 16. Ruling ids issued: 20.
A section that was missing. The NCAA men's lacrosse books 2022, 2024 and 2025 did not show section 6.5 as its own section, because the book prints "SECTION 5" without a full stop and the parser expected one. The parser now reads it. The section is one new row in each of three books, the 2022, 2024 and 2025 books (2 rows in the first two together and 1 in the third). In each, the text before it lost the part that belongs to 6.5. The new identifiers are listed in section 4. The two ruling records of the 2022 and 2024 books that sat under 6.4 but belong to 6.5 were moved there (see section 4).
Page headings and other text repairs.
- ATP Tour Rulebook 2023: page running heads that had been captured in the middle of rule text were removed. Rule rows changed: 88.
- MLB 2023: the book was re-read from the same source text. Page running heads were removed, hyphens were restored where a line break had split a hyphenated word, and a missing registered-trademark sign was restored. Rule rows changed: 58. Headings recovered: 4.
- PBA Tour Rule Book 2022, 2023 and 2024 (3 books): one row in each had its heading cut at the wrong place. Body texts changed: 3. Headings changed: 3.
- FA soccer (5 books: 2009, 2010-11, 2013, 2016 and 2017-18). In the 2009 book (
fa-soccer-2009) a new part, Fit and Proper Persons (s19b), was cut out of an existing part and its rows moved under it. Two rows that the book does not break were merged into the rows above them, which retired 2 permanent identifiers (section 4). Page banners at the end of rows were trimmed in these books. In all, 10 rows changed text. Rows added: 1. Rows retired: 2. - NFL 2024, 2025 and 2026 (3 books). In 31 sections the heading held the first words of the rule instead of the printed title. Each of these sections now has its printed title as heading, the rule words are back under it, and the printed title that stood glued to the end of the text before it was taken out. This repair changed 80 rows: 31 headings and 57 bodies. Small text faults were repaired in 5 rows of the three books: a number printed "4-1-2" stored as "41-2" and a word stored as "cointoss" in the 2024 book, and the list letters of rule 7.2.1, which stood in a group before their clauses, in all three books. In the 2024 and 2026 books the words of Rule 1, Section 4 (Chain Crew and Ball Crew), which were stored in Section 5, are back in Section 4 (4 rows). No row was added or retired. In all, 61 bodies and 31 headings changed in these three books.
- USGA golf 1997 (
usga-r-a-golf-1997, 1 book). A page is missing from the scan of this book: the page that carries most of Rule 1. 3 rows of Rule 1 were added from the words of Rule 1 that survive in the scan: Rule 1 (its heading only), rule 1-3 (its penalty sentence only) and rule 1-4 (complete). Together the three rows hold 53 words of body text. The 33 rows after them (Rules 2 to 34) moved down one place in the book's order; their text did not change. The new permanent identifiers are in section 4. What the scan lacks is still missing (see Known issues in this release).
4. Permanent identifiers
Rule rows. Since 2026.14, 2 permanent identifiers of rule rows were retired: fa-soccer-2009/s19.121, fa-soccer-2009/s19.137. Each points to the row it was merged into. In books that already existed in 2026.14, 7 identifiers were issued: usga-r-a-golf-1997/1, usga-r-a-golf-1997/1-3, usga-r-a-golf-1997/1-4, fa-soccer-2009/s19b, ncaa-men-s-lacrosse-2022/6.5, ncaa-men-s-lacrosse-2024/6.5, ncaa-men-s-lacrosse-2025/6.5. Of them, 3 are the new rows of Rule 1 of the golf 1997 book (section 3). The new books carry 23,123 newly issued identifiers. Of these, 1,268 are Premier League rows issued after the first load of their book (section 2), and 32 are the referee signals of NHL 2025-26, nhl-hockey-2025-26/29.4 to nhl-hockey-2025-26/29.35 (section 1; issued on 4 October 2026 in UTC, the time used in the lists). The deposit's list of retired identifiers (data/tombstones.json) holds 3,698 entries, against 3,696 in 2026.14. The lists are in changes/rule_ids_retired_since_old.csv and changes/rule_ids_issued_in_old_editions.csv.
Ruling records. Approved-ruling identifiers (<edition public id>/<path label>#ar<n>) are not part of the frozen public-identifier registry, and the ruling records are not part of this deposit. The repairs above retired 130 ruling identifiers and issued 178. The database held 2,524 ruling records at 2026.14 and holds 2,572 now. Retired ruling identifiers by cause. There are three causes: cross-references put back into rule text, the corrected end-of-example rule, and the move under section 6.5. Each line names its cause first:
- 31: cross-references put back into rule text, in NCAA football 2001 to 2005 (the 12 September 2026 split had cut them out as rulings).
- 13: cross-references put back into rule text, in NCAA men's lacrosse 2012 to 2024 (cut out as rulings by the same split).
- 2: cross-references put back into rule text, in NFL 2008 (cut out as rulings by the same split).
- 66: ruling records merged into the record printed above them when the end-of-example rule was corrected, in NCAA men's lacrosse 2012 to 2024.
- 16: ruling records merged into the record printed above them when the end-of-example rule was corrected, in NCAA women's bowling 2012 to 2024 (the same correction, applied to bowling).
- 2: ruling records moved under the new section 6.5 and renumbered, in NCAA men's lacrosse 2022 and 2024 (one record in each book).
Ruling identifiers issued, by cause. Each line names its cause first:
- 60: ruling records created by the corrected end-of-example rule, in NCAA men's lacrosse 2012 to 2024.
- 20: ruling records created by the corrected end-of-example rule, in NCAA women's bowling 2012 to 2024 (the same correction, applied to bowling).
- 63: ruling records created by the first split of the worked examples of NCAA men's lacrosse 2025 (edition 337).
- 33: ruling records of the worked examples of the new NCAA women's bowling 2026 book (edition 359).
- 2: the new identifiers of the two records moved under section 6.5 (NCAA men's lacrosse 2022 and 2024).
Apart from the two moved records, ruling records that survived keep their identifiers. The full lists are in changes/ruling_ids_retired.csv and changes/ruling_ids_new.csv, and the ruling counts that changed by book are in changes/ruling_rows_by_edition_changed.csv. One difference that needed an explanation: the NFL 2008 book had 77 ruling records in 2026.14 and has 75 now. Two records were cross-references and went back into rule text (the NFL 2008 line in the list of retired identifiers above).
5. The cross-reference table
The database has a table of cross-references, the places where one rule points to another rule. It was not in the 2026.14 database. It now holds 61,499 rows over 135 editions in 8 of 43 series, by series:
- NFL Football: 13,133 rows over 28 editions
- NCAA Football: 13,785 rows over 21 editions
- NCAA Basketball: 17,192 rows over 20 editions
- IFAB Soccer: 832 rows over 20 editions
- MLB Baseball: 4,339 rows over 14 editions
- NCAA Basketball (Women): 5,731 rows over 7 editions
- NCAA Track & Field and Cross Country: 5,080 rows over 17 editions
- NCAA Men's Lacrosse: 1,407 rows over 8 editions
Of the rows, 18,869 point to a rule row of the same book. This table is in the database. It is not in the deposited files. No file of this deposit carries it. In the covered series, 3 editions have no cross-reference rows yet: ncaa-basketball-2026-27, ncaa-basketball-women-2025-27, ncaa-football-2026.
6. Two changes to how the public reader shows text
These two changes are in the export program, which builds the files of this deposit. They change the reader files. They change no stored rule text and no word of the database.
The first-sentence rule. In the books whose text policy is truncated, the reader shows the first sentence of each entry. This release repairs entries that showed only a cut-off label instead of their first sentence. The cause was a list number or label at the start of the text (such as "1.", "Art 1." or a Roman numeral) that the export program read as the end of the sentence. 5,509 entries now show more than before. The release also shortens entries that showed more than one sentence, or that ended on the label of the next list item (for example "1." after a colon). 1,145 entries now show less. The rule is still not right for every entry. At least 425 entries fall short of it, and the list is not complete. The identifiers that were found and checked, with the kind of fault, are in changes/reader_known_residuals.csv. An entry that is cut earlier than its first sentence shows less text than the rule intends, never more. The list holds the entries that stop at a full stop inside a unit or a number (for example after "in." or "ft." in a measurement) as one kind of fault. Other entries of that sort exist and are not listed, for example a cut inside a word that is split across two lines or after a letter in a cross-reference. Rows shaped like a short label followed by a short title (NFL rows of the form "Item 1." with a title) show the label and its title and stop there. The books published in full are not affected.
The line-break fix. The printed line breaks of a book are stored in the database. The export program turned every line break into a new paragraph, so a sentence that runs over three printed lines appeared as three short paragraphs. The reader files now join a line break into a space when it falls inside a sentence: the line before ends with a letter, a digit or a comma, and the next line starts with a lower-case letter and does not open a list. The fix changes 78 books. Some wrapped lines stay separate on purpose, for example when the next line starts with a capital letter or a number, because no simple rule can tell those from real breaks.
7. Corrections to documents of release 2026.14
Four statements in the two deposited documents of release 2026.14 were wrong. The data of 2026.14 is not affected.
- The file
derived/README.mdnamed the database fingerprint860c3a7b9c27d457. That is the first cut of 15 September, which was withdrawn before deposit. The right fingerprint of release 2026.14 isc05eea6d34a1bee0, asderived/derived_provenance.jsonanddata/manifest.jsonsay. - The Known issues section of the 2026.14
README.mdsaid thatsource_document.source_urlwas filled for only 3 of the 537 registered documents (retrieval_datefor 6). Those are the figures of the withdrawn first cut. In the snapshot of 2026.14 the counts are 4 of 539 (retrieval_datefor 8). The corpus paragraph of the same README, with 539 documents, was right. - The Known issues section of the 2026.14
README.mdsaid that the two retired FA editions (fa-soccer-2014andfa-soccer-2009-10) "no longer appear indata/editions.jsonorreader/". They do appear. Each is listed indata/editions.jsonwithretiredset to true, and its reader file is a short redirect stub that names the edition it was merged into. Their rows are left out of the counts, and their identifiers are on the list of retired identifiers indata/tombstones.json. The export program does the same today. - The 2026.14
README.mdsaid, in its NFL re-extraction section and again in its known limits, that the NFL books of 1999 to 2005 have no worked examples in the database. That holds for 1999 to 2003 and for 2005. The 2004 book has worked examples.
Known limits carried into this release
- The 2013 NFL rule book has no PDF available in this database's source material.
- The 2011 NFL book's Approved Rulings casebook (pages 132-244 of that PDF) was not extracted; it is paginated separately from the rule text.
- The NFL books of 1999 to 2003 and 2005 have no worked examples in this database, by design (their own front matter states rulings are omitted). The 2004 book has them.
Also carried forward, unchanged from 2026.14
Every edition record in data/editions.json carries an explicit policy field naming its text-exposure tier. The tiering is unchanged from prior releases and applies to the exported editions.
Contents
- data/: catalog files (totals, sports, series, editions, manifest, unit types, concept schemes, tombstones)
- derived/: six aggregate tables (TSV files: plain text, columns separated by tabs, openable in a spreadsheet) with a provenance stamp naming the database snapshot they were generated from; no rule text. See
derived/README.md. - reader/: one JSON file (a plain-text data format that programs and most editors read) per exported edition, plus a short redirect stub for each retired edition. Each full file holds the node tree with headings, text per the exposure policy, and stable public identifiers of the form
<edition-slug>/<path-label> - changes/: lists of what changed since 2026.14. CSV files (plain text, columns separated by commas, openable in a spreadsheet; ids, labels and counts only, no rule text). Where a list gives "words", it is the exporter's token count of body text. One line each:
new_editions.csv: the new books, with sport, governing body, title, year, edition id, public id, rows and words.new_series.csv: the new series, with sport, governing body, number of books and rows.new_bodies.csv: the new governing bodies, with abbreviation, type and country scope.by_sport.csv: books, rows and words by sport, in 2026.14 and in this release.added_after_first_ingest_rows.csv: every row added after the first load of its book, with its public identifier and the time it was issued (UTC): the 1,268 Premier League rows (section 2) and the 32 referee signal rows of NHL 2025-26 (section 1).added_after_first_ingest_batches.csv: the same rows counted by book and by hour of issue.changed_editions.csv: for each older book that changed, the number of rows changed, added and retired, the words before and after, and the kind of repair.rule_ids_retired_since_old.csv: the rule identifiers retired since 2026.14, each with the identifier it points to and the reason.rule_ids_issued_in_old_editions.csv: the 7 identifiers issued since 2026.14 in books that already existed (section 4).ruling_ids_retired.csv: the retired identifiers of worked examples (approved rulings), with the cause.ruling_ids_new.csv: the new identifiers of worked examples, with the cause.ruling_rows_by_edition_changed.csv: the number of worked-example records per book, before and now, for the books where it changed.citation_by_series.csv: the cross-reference table counted by series (the table itself is not in this deposit).reader_known_residuals.csv: the entries found so far that fall short of the first-sentence rule of the reader files (section 6), each with the kind of fault. The list is not complete.
- docs/: the identifier-registry design (stability guarantees) and the rule-text exposure policy
- CITATION.cff: citation metadata
Text exposure policy
Rule text is tiered per edition: full (public domain, published up to 1930), truncated (first sentence per provision), or metadata only (bibliographic details, no text). This deposit contains exactly what the public website serves; enforcement happens at export. See docs/EXPOSURE-PROPOSAL.md. In this release 36 exported editions are full and 310 are truncated.
Identifier stability
Every rule node carries a public identifier frozen at first publication. Identifiers are never renumbered or reused; retirements are tombstoned: a retired identifier stays on a list of retired identifiers (data/tombstones.json), with its retirement date and, where there is one, the identifier it points to. See docs/IDENTIFIER-REGISTRY-BRIEF.md. A public identifier and the catalogued year of its edition are allowed to disagree by design: edition usa-pickleball-rulebook-2019 keeps its identifier and reader-file name although its catalogued year was corrected to 2020 on 2026-09-07.
Rights and reuse
This database has three layers, licensed separately. The compilation (corpus selection, parsing, hierarchy, unit taxonomy, metadata, annotations, and coding schemes) is released under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). You may share and adapt it for any purpose, as long as you credit it. If you reference it in published work, cite it using CITATION.cff or the DOI listed under Permanent identifiers below. The code (schema, exporter, example queries) is MIT. The rule text is the property of the respective governing bodies. It is not mine to license, and no license granted above extends to it. The public files show complete text only where the source is in the public domain (in this release, editions published up to 1930). For the other editions they show the first sentence of each provision, or bibliographic metadata only.
Known issues in this release
- Edition
fa-soccer-2014andfa-soccer-2009-10were retired on 2026-09-06 intofa-soccer-2013andfa-soccer-2009respectively (the merge that earlier releases flagged as an erratum). Each is still listed indata/editions.json, withretiredset to true. Its reader file is a short redirect stub that names the edition it was merged into. Its rows are left out of the exported counts. Its identifiers are tombstoned: they stay on the list of retired identifiers indata/tombstones.jsonand are never reused. - The source address field of the schema (
source_document.source_url) is filled for only 33 of the 607 registered documents. The retrieval date field is filled for 66. - FA Standard Code of Rules layout. In some of the older FA handbooks the parts that hold the Standard Code of Rules are stored as a flat run of rows. A numbered list that restarts inside one rule appears as separate rows with repeating labels, and the fee and pitch tables sit inside large rows with their cells in mixed order. No repair of this layout is part of this release.
- FA soccer list items whose text sits in the heading field. In the FA soccer handbooks, list items (for example those marked "(a)" or "2.4") keep their text in the heading field. This was checked for 354 rows in 13 books. In 326 of them the body holds only a short marker (2 to 9 characters, such as "and" or "or") and the item text is in the heading. In the other 28 the text is split between the two fields: the heading holds the start and the body holds 52 to 283 characters. The words are right but stored in the wrong field. The reader files publish headings in full. Other rows of the same kind may exist. No repair is part of this release.
- FA handbooks 2023-24 to 2025-26 are not in the database yet. The latest FA soccer handbook in this release is 2022.
- Golf 1997 (
usga-r-a-golf-1997) still lacks most of Rule 1. A page is missing from the scan of this book: the page that carries most of Rule 1. The database now has 3 rows of Rule 1, built from the few words that survive: Rule 1 with its heading only, rule 1-3 with its penalty sentence only, and rule 1-4. It has no rows for 3 labels that the 1996 and 1998-99 books have: rules 1-1 and 1-2, and the Wrong Ball definition. Rule 1-3 lacks the start of its text, and the same lost page also cut off the end of a note of the Water Hazard definition. Nothing was added from other editions. The reader files do not mark the three rows as partial. - Golf 2019 (
usga-r-a-golf-2019) has no separate rows for its middle-level rules. For example, rule 8.1b has rows for its numbered parts such as 8.1b(1), but no row of its own. The book therefore does not line up level by level with the 2023 book, and a few rows are out of reading order. - NFL text faults that remain. Two text faults were repaired in the 2024 book: a number printed "4-1-2" that was stored as "41-2", and a word stored as "cointoss" (now "coin-toss"). The same two faults are still in 4 rows of the 2025 and 2026 books (rules 3.36.1 and 4.2.1 of each book). The printed pages of those two books were read and show the right form. The word "cointoss" is also stored in 5 rows of older NFL books (
nfl-football-2018,nfl-football-2019,nfl-football-2020,nfl-football-2021,nfl-football-2023), which were not read against the printed pages. In the 2024 book the list letters of rules 12.1.3 and 15.3.1 are out of place. Other rows of the 2024 and 2025 books lost a hyphen at a line end, as "41-2" did (for example rules 3.17.7 and 8.3.2 of the 2024 book); they have not all been read against the printed pages. In the 2025 and 2026 books the heading of rule 4.7.4 is cut short. No repair of these rows is part of this release. - Premier League handbooks 2023-24 to 2025-26: heading letter and list letters. In the Commercial Rights Protocol the section heading "Process" has lost its letter "C." in each of the three books: the word stands at the end of clause 4 (
APP.fmv_commercial_rights.4), 3 rows. In the committee constitution, 9 rows hold a list letter that is stored twice, in the wrong place or missing: clauses 1 to 5 of the 2023-24 handbook, and clauses 7 and 11 of the 2024-25 and 2025-26 handbooks (Misc.PFNCC.1and so on). The reader files show only the first sentence of these rows. - NHL, older books: Rule 11.2 and the referee signals table. In the four older NHL books in the database (
nhl-hockey-2009-11,nhl-hockey-2015-16,nhl-hockey-2017-18,nhl-hockey-2018-19), Rule 11.2 still has a line break inside "non-compressed", and the referee signals table of Rule 29 is not repaired: only the 2025-26 book was. - The cross-reference table is not in this deposit. It is in the database only. In the database, the three newest NCAA editions of series that the table covers have no rows yet:
ncaa-basketball-2026-27,ncaa-basketball-women-2025-27,ncaa-football-2026. - USBC Playing Rules, chapters 7 and 8. In both USBC books (2025-26 and 2026-27) chapters 7 and 8 are stored as whole chapters, with no rows for the rules inside them (4 chapter rows).
- NCAA soccer 2026-27 and NCAA baseball 2025-26. The text of diagrams and tables sits inside rule rows in the order the extraction read it.
- NFL 2007. Some section titles are stored in upper case where the book prints title case (carried from 2026.14).
Files
pierce-2026.15.zip
Additional details
Related works
- Is supplement to
- Dataset: https://adamcopeland.co/rules/ (URL)
- Is supplemented by
- Software: https://github.com/adamcopeland-code/pierce-tools (URL)