Other pages in this series (34)
Lunar-eclipse articles: types and match rate, by source
'月食' articles classified by type (observation, forecast, ritual, non-occurrence, discussion, not an eclipse); the metric is the share of observation reports within ±1 day of a computed umbral eclipse visible at the site. Up to 1.1 only a lower bound without type classification was given.
Lunar-eclipse observation reports match at 71–79% in all three sources, a little below solar. Night events slip a day easily (after midnight), and Goryeosa's '月食 + star name' articles are occultations, not eclipses.
- Formula
- Per type: matched / dated articles. Lunar eclipses from skyfield's eclipselib, 1,535 umbral eclipses in 918–1910 (1,535 of the NASA catalog's 1,538 match within ±1 day); visibility is judged by the Moon's altitude at maximum eclipse. Window ±1 day.
- How to read it
- The observation-report value is the headline. Forecast, ritual and discussion rates differ by source (ritual: Ilgi 44%, Sillok 25%) and cannot be generalized as higher than for solar eclipses — read the per-type rows.
- Caveats
- Articles of the form '月食畢大星' (40 in the Goryeosa treatise) are occultations of a star by the Moon and were classified as 'not an eclipse' (the Jeolyo's one 'not an eclipse' row is an article punishing a missed solar-eclipse forecast; its '日月薄食' quotes an old rule). From 1.2.1 an editing rule places the reply dispatches sent to the Qing Board of Rites after an eclipse ('月食已過, 禮部回咨') under discussion; 1.2.2–1.2.3 added articles of the same kind whose quote lacks 回咨, read in the original (111 Ilgi and 1 Sillok articles; the coder's type is kept in type_coder). Abstinence (齋戒) and eclipse-day postponement articles coded forecast, 14 in all, move to ritual under the 1.2.2–1.2.3 rule. Eclipses after midnight are often dated to the previous day, hence the ±1-day window (after_midnight flag).
- Reliability card
- externalExternal match — compared with an astronomical computation
Goryeosa treatise: 153 of 196 observation reports within ±1 day = 78.1% (95% 71.8%–83.3%).
Sillok: 139 of 195 observation reports within ±1 day = 71.3% (95% 64.6%–77.2%).
Ilgi: 123 of 155 observation reports within ±1 day = 79.4% (95% 72.3%–85.0%).
Classification by type: 1,534 articles; 150 blind re-coded — coder-to-coder agreement 82.7%, κ = 0.78 (pooled κ = 0.94); κ = 0.91 once the editing rules (回咨, 齋戒; 1.2.1–1.2.3) are applied to both sides.
| Source | Type | Articles | Dated articles | Match ±1 d | Share | 95% lo | 95% hi | Same day |
|---|---|---|---|---|---|---|---|---|
| Goryeosa treatise | discussion | 1 | 1 | 0 | 0.0% | 0.0% | 79.3% | 0.0% |
| Goryeosa treatise | not occurred | 21 | 21 | 14 | 66.7% | 45.4% | 82.8% | 42.9% |
| Goryeosa treatise | observation | 201 | 196 | 153 | 78.1% | 71.8% | 83.3% | 44.4% |
| Goryeosa treatise | observation (day level) | 196 | 196 | 153 | 78.1% | 71.8% | 83.3% | |
| Goryeosa treatise | not an eclipse | 40 | 39 | 0 | 0.0% | 0.0% | 9.0% | 0.0% |
| Jeolyo | observation (day level) | 0 | 0 | 0 | ||||
| Jeolyo | not an eclipse | 1 | 1 | 0 | 0.0% | 0.0% | 79.3% | 0.0% |
| Sillok | discussion | 62 | 53 | 7 | 13.2% | 6.5% | 24.8% | 5.7% |
| Sillok | forecast | 54 | 54 | 11 | 20.4% | 11.8% | 32.9% | 3.7% |
| Sillok | not occurred | 13 | 13 | 6 | 46.2% | 23.2% | 70.9% | 15.4% |
| Sillok | observation | 195 | 195 | 139 | 71.3% | 64.6% | 77.2% | 45.1% |
| Sillok | observation (day level) | 195 | 190 | 136 | 71.6% | 64.8% | 77.5% | |
| Sillok | not an eclipse | 28 | 28 | 0 | 0.0% | 0.0% | 12.1% | 0.0% |
| Sillok | ritual | 16 | 16 | 4 | 25.0% | 10.2% | 49.5% | 12.5% |
| Ilgi | discussion | 269 | 269 | 60 | 22.3% | 17.7% | 27.6% | 5.2% |
| Ilgi | forecast | 82 | 82 | 8 | 9.8% | 5.0% | 18.1% | 4.9% |
| Ilgi | not occurred | 6 | 6 | 3 | 50.0% | 18.8% | 81.2% | 16.7% |
| Ilgi | observation | 155 | 155 | 123 | 79.4% | 72.3% | 85.0% | 40.6% |
| Ilgi | observation (day level) | 155 | 153 | 122 | 79.7% | 72.7% | 85.3% | |
| Ilgi | not an eclipse | 49 | 49 | 1 | 2.0% | 0.4% | 10.7% | 2.0% |
| Ilgi | ritual | 341 | 341 | 146 | 42.8% | 37.7% | 48.1% | 23.8% |