Every ratified world record in athletics' twenty standard events, men's and women's, as one progression table: 1,837 records across all 40 event-sex progressions, from the amateur miling of the 1850s to this year, when history moved twice (Josh Kerr took the mile record El Guerrouj had held for 27 years, and Sabastian Sawe's 1:59:30 in April became the first ratified sub-two-hour marathon; both are rows here and both are pinned by test). Marks parse to seconds for track and road and metres for field, surviving the source's whole zoo: rowspanned cells where one hand-timed 10.6 spans a dozen sprinters and Bolt's name spans his 9.69 and 9.58, the European dot era that wrote 3:57.0 as 3.57.0, ISO dates on some pages and spelled months on others, and auxiliary indoor, unofficial and annulled tables excluded by their captions and headings so Kerr's indoor mile does not impersonate the outdoor record. Days_Stood computes inside each progression with the current holder flagged: Bolt's 9.58 is into its 18th year, and the Soviet-era field records (Kratochvilova's 800m, Koch's 400m, Sedykh's hammer) all stand past 14,000 days, the doping-era conversation as a sortable column. Monotonicity is enforced by test with tolerance only for the hand-to-electronic timing seams. Built with a 12-test acceptance suite: one current record per progression, ordered dates, Bolt, Kerr's new mile, the sub-two marathon and the immortal Soviet marks.
1,837 rows in one CSV file, with 12 columns in total: Event, Sex, Type, Mark, Mark_Numeric, Unit, Athlete, Nation, Record_Date, Days_Stood, Current, Event_URL.
It is built by programmatically scraping en.wikipedia.org, last pulled on 2026-09-23. Datasets are versioned; older versions stay downloadable.
Yes. Every CodeSights dataset is completely free as a CSV download; a free account is all it takes. Anyone can preview the data without signing in.
Yes. The exact Python scraper that built it is viewable on the dataset page by any signed-in member, so every number is reproducible.
Automated scraping leaves room for error and the underlying sources change over time, so no version is guaranteed accurate or complete. If a number matters, verify it against the original source.