Every solar eclipse from 1601 through 2200, with columns for what else the world was doing while the sky went dark: 1,391 eclipses, 1,019 past and 372 still to come, because eclipses are computed rather than observed and the future rows are as solid as the past ones (the longest totality in the file, over 7 minutes, falls in 2186 and has not happened yet, honestly flagged Is_Future). Each row carries the canon from Wikipedia's per-century tables following NASA's eclipse catalog: date, type (497 partial, 448 annular, 373 total, 73 hybrid), saros cycle, gamma, magnitude, central duration with a seconds twin (a hybrid at its transition can genuinely read 0m 00s), the point of greatest eclipse as coordinates, and the visibility path. The twist columns cross-reference all 366 of Wikipedia's Month-Day pages, whose Events, Births and Deaths sections record history by exact calendar date: Same_Day_Events, Same_Day_Event_Count, Born_That_Day and Died_That_Day. The dataset found its own headlines: the 1879 annular eclipse row carries the Battle of Isandlwana, fought under what the Zulu army called the day of the dead moon, and the 1919 row records Arthur Eddington testing Einstein's general relativity beneath that very eclipse. About a quarter of 20th-century eclipse days carry a recorded event; most days in history, even eclipse days, made no famous news, and the future rows are honest NAs because nothing has happened on those days yet. Built with a 14-test acceptance suite anchored on Eddington, the Great American Eclipse and the not-yet-happened record holder.
1,391 rows in one CSV file, with 19 columns in total: Eclipse, Date, Year, Century, Type, Saros, Gamma, Magnitude, Central_Duration, Central_Duration_Seconds, Latitude, Longitude, Visible_In, Is_Future, Same_Day_Events, Same_Day_Event_Count, Born_That_Day, Died_That_Day, Eclipse_URL.
It is built by programmatically scraping en.wikipedia.org, last pulled on 2026-09-12. 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.