Fonti
Questa pagina dedica una riga a ogni quantità che il motore calcola, e ogni riga dice tre cose: da che cosa il motore la ricava, rispetto a che cosa è stata verificata, e su quale rung della scala qui sotto quella verifica la colloca.
La scala delle prove
Le quantità sono state verificate con riferimenti di solidità diversa, e sapere da dove viene una quantità è il primo passo per soppesarla accanto alla sua vicina. I rung qui sotto sono il secondo: vanno dal più forte al più debole, e accanto a ciascuno c’è il numero di quantità che vi stanno.
- 0 Misurata 5 nel registro
- Una risposta delle effemeridi: si interroga il cielo e la risposta si legge. Non è affatto una regola tramandata, ed è per questo che sta sopra i rung sottostanti e non su di essi.
- 1 Un riferimento eseguibile 13 nel registro
- Un altro programma la calcola, indipendentemente da questo, ed è stato eseguito sull’intero dominio anziché a campione. Dove la materia è una tradizione controversa questo significa coerente con un’implementazione diffusa, mai verificata.
- 2 Due testi che concordano 13 nel registro
- Due fonti tramandate, indipendenti l’una dall’altra, che nominano la stessa cosa allo stesso modo. È lo standard per tutto ciò che non si può derivare.
- 3 Sovradeterminazione 9 nel registro
- Una struttura con più vincoli che libertà, per cui una risposta sbagliata rompe molte cose insieme e quella giusta non ne rompe nessuna.
- 4 Un testo che verifica sé stesso 10 nel registro
- Un testimone solo, esaustivo o ridondante su ciò che è in questione, così da poter essere tenuto al proprio stesso resoconto.
- 5 Un testo, non verificato 11 nel registro
- Una fonte sola, senza nulla di disponibile su cui verificarla.
- - Nulla registrato 1 nel registro
- Il motore porta la quantità e nessuna fonte la sostiene. Dove accade lo dice anche il disegno: quei nomi restano senza colore finché una non arriva.
Un rung non è un verdetto. Dice che cosa potrebbe andare storto senza che nessuno se ne accorga, che è cosa diversa da quanto storto sia qualcosa: il telaio delle ventotto dimore è retto da sovradeterminazione ed è esatto a un settantesimo della domanda più stretta che gli si ponga mai. E un rung non è nemmeno una proprietà della quantità: si sposta quando si sposta lo scaffale, in entrambe le direzioni, e una fonte che si limita a confermare lo sposta quanto una che contraddice.
La tavola qui sotto è citata, non tradotta. È un registro di citazioni (le quantità come le nomina il motore, edizioni, capitoli, i programmi su cui ciascuna è stata misurata e gli intervalli su cui sono stati eseguiti) e questo progetto lo tiene in inglese, la lingua del suo codice sorgente. Ciò che è tradotto è tutto quello da cui si decide: i rung qui sopra, che cosa significa ciascuno, e che cos’è ogni strato.
La colonna della scuola dice sotto quale valore dichiarato sta una quantità, nelle parole con cui lo si passa: huoling: fixed, yuan: futou. Un trattino dice che la quantità è portata come propria della tradizione e non come la lettura di una parte, ed è un’affermazione, non una casella lasciata vuota. Di chi sia una regola e quanto sia attestata sono domande diverse, perciò le due colonne stanno accanto e nessuna delle due sposta l’altra.
Lo strato calendariale
| Quantità | Scuola | Rung | Si regge su | Verificata su |
|---|---|---|---|---|
| the twenty-four solar terms | — | 0 | Swiss Ephemeris: the Sun's apparent longitude crossing a multiple of 15°, aberration and nutation included | lunar-javascript, 1926 dates from 1902 to 2098, agreeing on every one |
| the lunar date | — | 0 | the new moon on 120°E, the Moon's elongation from the Sun driven to zero | lunar-javascript, the same 1926 dates |
| the four pillars | — | 1 | the sexagenary count anchored on the terms and on the day boundary | lunar-javascript, year, month, day and hour on all 1926 dates |
| the true-solar correction | — | 0 | Swiss Ephemeris's lmt_to_lat for the equation of time, plus the offset from the zone's meridian | nothing independent: the ephemeris is the instrument here and not a witness |
| 納音, hidden stems, ten gods, twelve stages, void branches, the luck cycles | — | 1 | transmitted tables and derivations over the pillars | lunar-javascript, 479 charts over the same two centuries, the eleven disagreements all inside China's moving clocks |
Qí Mén Dùn Jiǎ 奇門遁甲
| Quantità | Scuola | Rung | Si regge su | Verificata su |
|---|---|---|---|---|
| the earth plate | — | 1 | counting the instruments and the marvels through the Luoshu | all eighteen published arrangements reproduced without a cell out of place, and qimen-dunjia over 160 charts |
| the ju under 茅山, the yuan from the term’s first instant | method: maoshan | 1 | the term in force, cut into sixty-shichen blocks from the instant it began, the third running to the term’s end | qimen-dunjia 2.1.0, 160 charts from 2000 to 2023, all thirteen quantities — and that reference computes 茅山, which is what this row used to call 拆補. Stated in print by 劉文元《奇門啟悟》2008 pp. 33-34 with both edge cases numbered, and by 唐頤《圖解奇門遁甲大全》2009 pp. 127-128 with the same two as 取 and 舍. The term-to-局 table under it is confirmed entire in print: 《奇門遁甲金鏡寶鑑》 heads each of its eighteen 局 with the four term-and-yuan slots that take it, and all seventy-two cells agree; 《御定奇門寶鑑》卷二 prints the same table as a verse for both dun |
| the ju under 拆補, the yuan from the 符頭 | method: chaibu | 1 | the day pillar's place in the fifteen-day cycle headed by 甲 and 己 | kinqimen over 3652 days, and fengshui-hacks on 260 of 266 moments, the six being 超神 windows |
| the ju under 置閏 | method: zhirun | 4 | 《奇門遁甲統宗》卷一 置閏法: whole fifteen-day blocks, the drift paid off by a repeated 芒種 or 大雪 | the text's own worked example over 1717–1719, every date and every dated term reproduced; ktonko.com confirms the structure piecewise. **Where the repeated block goes is contested**: 《奇門遁甲金鏡寶鑑》卷之一 repeats the term the year's leap month falls under instead, works it over 1678 and 1691 — both leap months confirmed here — and rejects the solstice placement outright. 《遁甲集成》第四冊 prints this passage a second time, Kangxi example and 甲午 slip and all, inside a work catalogued as 《奇門寶鑑》 — a collation and not a second witness, so the count is unchanged. Its 超接訣 leaf adds the rule's own ground: 「於逐年芒種節上，於大雪前尋之。蓋緣芒種節後是夏至…大雪之後是冬至」, the leap sitting at the two terms the polarity turns after **A second work states the same rule and derives its threshold**: 《奇門探索錄》卷三 pp. 21–22 gives 「置閏必以九日起數…八日以前得一氣之先，不必置閏。八日以後交九日，居一氣之末，故須置閏」 and 「超法自一日至十一日，接法自六日至一日」, which is `MAX_CHAOSHEN` = 8 exactly, and works the rule twice over 萬曆己丑 (1589) — eleven dated blocks, the leap flag and the day the intercalation stops all reproduced here. **The rung does not move**: its statement of the rule is a redaction of the 統宗's own sentence and so is not a second independent witness to it, and the two count the threshold a day apart in the same words **A third text prints both placements as one rule**: 《金函玉鏡奇門遁甲秘笈全書》上冊 p. 87 超神接氣法 — 「若其年閏正、二、三、四月者，必閏大雪一氣之候；如閏五、六、七、八、九月者，必閏芒種一氣之候」 — so the block is always one of the 統宗's two terms and the leap month chooses between them, which is 《金鏡寶鑑》's criterion. One witness, so it moves no value; it changes what the two declared ones are. **And the same work's other volume names the pin and derives the rule**: 卷二十一 p. 425 超神接氣置閏訣 reads 「冬夏二至前起閏…以三十日零五時二刻為率，以三十日分六局，以餘五時二刻置閏，遇芒種大雪超過九日便當置閏」, which is `MAX_CHAOSHEN` = 8, the solstitial placement, and the intercalation derived from the length of a solar month. So the fourteen-day 訣例 its p. 87 reports is one it names and does not use. Read on the plate 2026-09-09 **A fourth text names a third placement and a party with no leap at all**: 《奇門遁甲折衷》 ([清] 甘國壁) calls the solstitial rule 諸書's — 「其閏法定以大雪芒種兩氣中為限」 — and puts the block at any of the eight 節 instead, 「二至二分及四立之前，曆書謂之四絕四離，遇有積餘皆可置閏。惟於八際置閏」, while the 耕山陳氏 it quotes holds 「奇門遁甲惟有超接之機，而無置閏之法」. Four readings inside one work quoting three others, so the rung does not move; what it adds is that the solstitial placement is attributed to nobody by a second work looking for its author. Read on the plate 2026-09-10 |
| the heaven plate, the nine stars, the eight gates, the eight spirits, 值符, 值使, 旬首, 空亡, 驛馬 | spirits: dun | 1 | the turn of the plate until the instrument stands over the hour's stem | fengshui-hacks, 260 charts over the eight outer palaces — it leaves the centre starless, so it cannot weigh the ninth; and qimen-dunjia over 666 charts agreeing on the earth plate, the heaven plate and all nine stars, including the 147 that put a stem in the centre. 《奇門遁甲全局》 in print, four charts of 陽遁一局 and 陰遁一局 read on the plate — the yin board agrees on all eight spirits, the yang board keeps 白虎 where this engine renames it. 《奇門遁甲金鏡寶鑑》 in print, the 起星訣 and 起門訣 of 陽遁一局 read whole — all six 直符 stars and all six 直使 gates agree by name, and all twenty palace disagreements are the centre counted as a station, one of them carrying a cut slip on top of it — which is `centreTravel`, computing both values and moving no plate under either. Its prose, swept whole for 卷一~卷三, pairs the middle seats as this engine does — 勾陳 with 白虎, 朱雀 with 玄武 — and selects between the two names by the matter asked rather than by the dun. 《御定奇門寶鑑》 in print states 值符隨時干 and 值使隨時支 in the engine's own terms, lodges the centre's decade out to 坤二 — the counter-witness to the palace-5 reading — and pairs the middle seats 勾陳/白虎 and 朱雀/玄武 while renaming neither. 《奇門探索錄》卷三 p. 22 lays a whole board in prose — 陰遁四局上元 for 萬曆己丑十一月初六日庚戌日戊寅時 — and it agrees on all nine earth-plate stems, on 旬首, 值符 and 值使, on all eight gates by palace, on all eight spirits by palace and on the 門迫 at 坎; it keeps 勾陳 and 朱雀 in both dun, which is the 御定 reading in a second work, and it parts on two stars its own turn contradicts (a slip) and on one heaven-plate cell, the 乙 lodged out of the centre travelling to 兌 where this engine leaves it in the fifth |
| 門迫 | — | 3 | the gate's phase controlling the palace's, which is one of the five relations | 《統宗》卷一 迫 prints eleven of the thirteen cells with its own reasons beside them; the two it omits its own reasons cover |
| 八門宣明克應動靜占驗 — the hexagram of a gate over a palace | — | 4 | the 周易's own arrangement of the sixty-four, read with the gate's trigram above and the palace's below | the arrangement is closed and checks itself — sixty-four cells, each name once, the eight doubled trigrams on the diagonal, and every figure the number's own at U+4DC0. What the chapter adds is the orientation, and it stands on 《奇門探索錄》卷十一 八門宣明克應動靜占驗, sheets 189–216, which prints all sixty-four — eight sections, one to a gate, each run over every palace. Twelve cells read on the plate — 杜加坎 渙, 杜加巽 巽, 杜加離 家人, 杜加坤 觀, 景加坤 晉, 死加巽 升, 死加坎 師, 死加兌 臨, 死加乾 泰, 開加休 訟, 開加艮 遯, 開加巽 姤 — which agree with each other and with no second work, none on this shelf pairing the gates this way at all. Asserted cell by cell in `test/hexagram.test.ts` |
| 五不遇時 | — | 3 | the hour's stem controlling the day's at the same polarity | the ten transmitted pairings — and the derivation found the mnemonic incomplete, the condition striking twice on two days in ten; and 《奇門遁甲金鏡寶鑑》卷之四 states the rule with its condition in print — 「時干剋日干。陽忌陽，陰忌陰」 — so the polarity clause is transmitted and not only inferred; and 《奇門探索錄》卷三 p. 52 states the rule bare — 「時干剋日干也，甲日庚時、乙日辛時之類」 — reaching for the same two same-polarity instances the 金鏡寶鑑 uses to illustrate the clause, so the pair is the tradition's and one work spells out what the other leaves to it |
| 驛馬 | — | 3 | the branch facing the triad's 長生 | 申子辰馬在寅 and its three fellows |
| 旺相休囚死, 門宮, 星宮 | strengths: season | 3 | the five-phase reckoning against the season and against the palace | stateable in a sentence and asserted against it; 門迫 is one of the five relations and a test holds the two to each other. **Contradicted in print**: 《奇門遁甲金鏡寶鑑》卷之四 states and tabulates the same five relations read outward from the star, which transposes 相 with 休 and 囚 with 死, and extends them to the day and hour where this engine reads the month alone |
| 入墓 and 六儀擊刑 | — | 5 | transmitted tables, held and deliberately not derived: 入墓 parts from the twelve stages | 《統宗》卷一 奇門四十格 in print — 擊刑 decade for decade and palace for palace, 入墓 three of its four, 戊 in Qian having no line of its own; 《奇門遁甲金鏡寶鑑》卷之四 prints the 擊刑 palaces of all six decades and agrees cell for cell, which is a second print for that half of the row and none for 入墓 |
| 十干克應 — the fourteen pairings shipped | — | 2 | the 四十格 verse and its tabular redactions | seven witnesses collated cell by cell, and the collation cut them to four lines: 《金函玉鏡》卷二 prints all 81 and B, H and the two Chinese pages are its own line in four states. Eleven pairings pinned to a line of the verse, one of them — 上格 — to a line the recensions divide over and now with a second print behind it. **戰格 shipped as 太白同宮**: every table names the cell so, and 《神奇之門》 states 「名為太白同宮，又名戰格」, which makes 戰格 the alias this file had been carrying as the name. **Two more shipped on J and 《元靈經》 together**: 青龍耀明, which 《開悟之門》 grades 第三吉格 and 《神奇之門》 seconds, and 日奇入霧, where both lines give a head verdict. The other sixty-seven are refused. 青龍華蓋 is named by four lines and given a fortune by none — the cell to reach for if a fourth valence is ever argued — and 戊 over 戊 collides with the board's own 伏吟. **And the best-titled candidate left on the shelf is a homonym**: 《奇門探索錄》卷四 p. 64 十干克應訣 is ten single stems read off the hour — 六甲天福, 六丁玉女, 六庚天刑 — each with an omen and a term, not eighty-one pairs. Established on the plate 2026-09-09 |
| 八門 — what each gate is chosen for | — | 2 | 《太乙金鏡式經》's domains and the 執事歌's errands | a Tang witness, two independent redactions of the verse, and 《統宗》卷二's prose table; the 統宗 passage confirmed on a second printed edition |
Dà Liù Rén 大六壬
| Quantità | Scuola | Rung | Si regge su | Verificata su |
|---|---|---|---|---|
| the four courses, 月將加時 and the 寄宮 table | — | 1 | the general set on the palace of the hour, every branch a constant step from where it lives | kinliuren over 17280 term-keyed boards: the four courses 17280/17280, the 上剋下 · 下賊上 marking 17280/17280 |
| the three transmissions and the 九宗門 | — | 1 | the nine rules, stated line by line in 《六壬大全》卷一 入手法 | liuren-ts-lib over the whole 8640-board space, 99.6 %, and 99.7 % where the two references agree — the two agreeing with each other on only 82.4 % |
| 返吟 | — | 4 | 《六壬大全》卷一's own statement of the exception, which is exhaustive about itself | no reference could check it; the text can be held to it and is |
| the five-phase assignments of the 十二天將 | — | - | the tradition transmits them and the engine carries none | no source is registered here, which is why the drawing leaves them in neutral ink |
| the 貴人 seat, by day and by night | guiren: chou | 2 | the 貴人歌 — 甲戊庚牛羊, 乙己鼠猴鄉, 丙丁豬雞位, 壬癸蛇兔藏, 六辛逢馬虎 | 《六壬大全》, which prints the verse with the branches interlined, against 《大六壬精解》 p. 26, which prints it with a prose gloss. The two disagree on one character — 六壬 for 六辛 in the 精校本 — and the interlinear 午寅 settles it, 壬 having been placed at 巳卯 two lines above |
| which hours are 晝 and which 夜 | zhouye: branch | 5 | 《大六壬精解》 p. 26: 卯辰巳午未申 are 晝時, 酉戌亥子丑寅 are 夜時, with a worked example on 甲子日 | nothing. 《六壬大全》 carries the 貴人 verse and the 晝順夜逆 rule and nowhere says which hours are 晝, so the division the engine computes rests on this one book |
Tài Yǐ Shén Shù 太乙神數
| Quantità | Scuola | Rung | Si regge su | Verificata su |
|---|---|---|---|---|
| the epoch — 上元積年 | epoch: jinjing | 2 | 《太乙金鏡式經》's own figure at 開元十二年甲子, 724 CE | the text states three reckonings differing by millions and all three are congruent modulo 周紀法 三百六十, which is the one residue every placement reduces by; the anchor is checked four ways inside the text. 《統宗》's own 公元基數 10 153 917 differs from it by 22 826 × 360 and 342 390 × 24, so a second lineage lays the identical board, and 《齊書》's eleven dated positions discard the third reckoning in circulation, 《淘金》's, which is out by a whole 元. The 年計 constants beside it in 卷一 — 周紀法三百六十, 元法七十二, 太乙小周法二十四, 天目周法十八, 紀法六十 — divide into one another and pass every check, which is what shows the broken constants of the other three registers to be local corruption and not a misreading |
| the placements — 太乙宮, 天目, 主算, 客目, 客算, 計神 | — | 2 | 卷三's procedure, run over the 年計 register | the text's own 立成, 144 rows and 864 cells: 850 exact, thirteen of the fourteen divergences settled by the text against itself, one written down unsettled. And from outside the manuals: the eleven dated positions of 《齊書·武帝紀》, all eleven reproduced with the 太歲 of each agreeing with the reign year that names it, plus five of 盧央's six worked boards recomputing in every printed value |
| 大將 and 參將 | — | 2 | the worked boards of 卷一, 卷六 and 卷九, which name their own 局, and 《統宗》's rule in words — 大將宮數 × 3 mod 10 — per 中國古代星占學 p. 476 | the fourteen worked instances are checked individually and every one of them is ×3 mod 10, which is the second witness stating what the first only exemplifies. Where they parted was the centre, and the stated rule is what the engine follows there: a count ending in five seats both generals in 五宮, which the second witness names 杜塞 and this engine leaves unnamed, the sentence saying what it is being a modern scholar's |
| the 太歲 the count yields | — | 1 | the count from the anchor year, one a year in either direction | yearGanzhi over four centuries, which is the pillar every almanac prints and the one check on this board that runs rather than being read |
| 合神 — the 六合 of the 太歲 | — | 4 | the 六合 pairing, each pair summing to thirteen on the wheel of branches | 卷一 states it by example — 「假令太嵗在子，合神在丑」 — and its worked board at 推六紀月建法 prints it beside the 計神 for a 局 this engine reproduces, 「太乙在一宫，武徳為天目，計神寅，合神丑」, which is the text checking itself |
| 八門直使 — one gate every thirty years | — | 4 | 卷四's procedure over the 小遊 count, 二百四十 to the circuit and thirty years to a gate, the ring opening at 開門 | 卷一 works it on two dates — 開元十二年甲子 standing on 開門 and 三十一年甲午 on 休門 — and the count gives both, which is the text checking itself about the thing in question |
| 五福太乙 | — | 2 | 卷五 推五福太乙法 on the 一萬三千三百三十一 count: forty-five years to a palace over the four corners and the centre, 二百二十五年 to the circuit | the text works it on a date — 「今開元十二年甲子在遼東十一年也」, the eleventh year in 黃始宮 — and that is what the count gives; and 周琮's memorial of 1071 in 《宋史·禮志六》 dates the body into 巽 in 984 and 坤 in 1029 by the temples the Song court built, asking for a third because it enters the centre in 1074, which the count also gives |
| 大遊太乙 | — | 5 | 卷五 推大遊太乙 on the same 一萬三千三百三十一 count: thirty-six years to a palace, 二百八十八年 to the circuit, opening from 七宮 | nothing. It is not one of the six placements the 立成 prints, and the text works no dated example for it; what it shares with 天目 is the walk and not the count |
| 三基 — 君基, 臣基, 民基 | — | 2 | 卷五's opening count, 二十八萬五千一十一, over the twelve fiefs from 戌 — thirty years to a fief for the sovereign, three for the minister, one for the people | nothing in 《金鏡》: 二十八萬五千一十一 and the 六百十一 printed beside it agree mod 三百六十, which checks the count and not what is done with it. From outside: 盧央 dates all three — 君基 into 午邦 in 714, 臣基 into 午邦 in 642, 民基 into 未邦 in 627 — and runs 《統宗》's arrangement beside 《金鏡》's for the first two, where a start from 午邦 and a 邦盈差 of 二百五十 compensate and land on the same fief |
| the conditions of 卷三 — 掩, 擊, 迫, 關 | — | 2 | 卷三's triggers, read over the placements the 立成 weighs | 卷三 states each condition three times and the three are three kinds of sentence, so the repetition fixes the wording and not the trigger. From outside: four of 盧央's worked boards name these four where this engine names them — 掩 at 207 BCE, 外辰擊 at 124, three 迫 at once at 210 BCE, 關 at 830 — and 迫's four distances are graded there in the same words |
| the conditions of 卷三 — 囚, 格, 對, whose subjects two lineages dispute | — | 5 | 卷三's triggers, read over the placements the 立成 weighs | nothing runs, and a second lineage disagrees about which bodies enter each. Two triggers had to be read here — whose bodies 格 admits, and whether 囚's 俱 is collective — and both readings are written down; 《統宗》 per 盧央 marks 格 on the host's 大將 as well as the guest's, admits a 沖 of branches where this engine tests the facing palace, and reads 文昌 in 太乙's palace as 關 or 掩 where 卷三's list puts it in 囚 |
| the nine palaces, numbered one seat off the 洛書 | — | 2 | 卷二 推九宮所主法, which gives the whole map — 「一宫在乾…二宫在離…三宫在艮…四宫在震…六宫在兑…七宫在坤…八宫在坎…九宫在巽」 — and the reason, 「所以差一宫以就乾位」 | 《統宗》 names the same seats, per 中國古代星占學 圖 5-6 and its worked board for 940 where 太乙 in 一宮 is glossed 乾; and the 立成 of 864 cells reproduces under this numbering and no other, since every 算 is a sum of the palace numbers walked |
| the element of each of the sixteen seats | — | 4 | 卷六, which states it for thirteen of the sixteen in one passage — 「假今髙叢木，以吕申、大炅同類為旺…」 — every one of them the element of the branch or trigram the seat stands on | the thirteen constrain the rule and the remaining three follow from it rather than being chosen; the drawing tints eight palaces by the seat each stands on, so a wrong element would be visible and is not checkable against a second witness |
Qī Zhèng Sì Yú 七政四餘
| Quantità | Scuola | Rung | Si regge su | Verificata su |
|---|---|---|---|---|
| the places of the seven governors | — | 0 | Swiss Ephemeris, the library the solar terms already stand on | the positions are the same tier-1 material as the terms; every difficulty on this board is in the frame and not in them |
| the frame — where the twenty-eight 宿 begin | xiudu: juxing | 3 | the 距星 themselves, placed at the instant of the chart with precession and proper motion in them, from sefstars.txt | over-determined: twenty-eight transmitted widths, a ring that must close on 360°, and 儀象考成's places agreeing to about a minute of arc per star. 《曆法通志》 pp. 273-275 print the 黃道 widths of eight 曆 together, none with an epoch, and record that the 授時 took the 紀元曆's and that such tables are computed from the 赤道 by proportion rather than measured — which is this section's argument made from the other side |
| 觜 and 參, the pair a degree apart | — | 3 | the same frame, at its narrowest lodge | 儀象考成 1744 read against the computed places: 0.5″ for 斗宿一, 7.6″ for 參宿一, 50.5″ for 觜宿一 — a seventieth of the tightest question the frame answers |
| 四餘 — 羅睺, 計都, 月孛 | luohou: descending | 5 | the mean elements the tradition transmits for them, 隱曜 being positions and not bodies | 《星度指南》's worked board, which fixes a palace and not a degree, ±15° |
| 紫氣 — the fourth 餘, placed to a palace and to no degree | ziqi: yinianyisu | 5 | 《張果星宗》's 大數 10228, one circuit in twenty-eight years — 一年一宿 stated arithmetically — anchored on 《星度指南》's dated board | nothing weighs the constant itself, and each half of the placement is one witness: that work's own 總論 gives twenty-nine years two columns from the 算法's twenty-eight, and the anchor is a single 1886 chart good to a palace, ±15°. The two siblings the same table carries err by 0.01 % and 0.20 % against the sky, and nothing establishes that this one is no worse |
| 命宮 by 立命 加時 | minggong: yuejiang | 3 | 「以生時加在太陽所在之宮，順數至卯」, which the texts state in one line | checked against the sky and not against a worked example: 卯時 returns the Sun's own palace, 酉時 the opposite, 午時 the palace ninety degrees along. A rule laid the other way fails all three |
| the twelve 人事宮 | — | 3 | counting from the 命宮, in a direction no source consulted states in terms another could be held against | over-determination: ten of the twelve names fail a reversal, the twelve are the Hellenistic houses in the Hellenistic order and not 紫微斗數's, and the transmitted 運限 order runs the same way |
Zǐ Wēi Dǒu Shù 紫微斗數
| Quantità | Scuola | Rung | Si regge su | Verificata su |
|---|---|---|---|---|
| the twelve palaces, the bureau, 紫微 and the fourteen main stars | — | 1 | 《紫微斗數全書》卷二's verses, counted from the lunar month, the hour and the day | iztro and fortel-ziweidoushu over 544 births from 1930 to 2020: fifty-nine of sixty-three quantities agreeing in every chart; and the 紫微 grids read off the Ming 南陽堂 plate and collated cell by cell — 150 of 150, including the two cells the first edition lost and an arithmetic had been standing in for; and both of those cells printed again by 《紫微斗數捷覽》 (明刊孤本, 1581), which is a different work of this transmission rather than a fourth dress of this book |
| the brightness grades | — | 3 | the table of seven grades closing 卷二 | the grid counted itself — eighteen stars in all twelve branches and three in exactly eight, which is what 祿存 following the year stem forces; 卷三's coarse summaries agree where they overlap |
| 天刑 and 天姚 | — | 2 | 卷二, 酉 and 丑 counted forward from the first month | the only placements 《全書》 shares with the 十八飛星 transmission, agreeing character for character across two lineages that agree about nothing else; 《紫微斗數捷覽》 is not a third witness of that crossing, being this transmission's own |
| 火星 and 鈴星 | huoling: fixed | 4 | 卷二's seats from the birth year's triplicity, the hour nowhere in the verse | both editions print the same rule and the Ming 南陽堂 recension a third time, which collates the line and adds no witness to it; agreement with the references is exactly the quarter of the sample whose hour offset is zero, which is what names the divergence rather than hiding it |
| 解神 | — | 4 | 「解神從戌上起子，逆數至當生年太歲上是也」, placed off the birth year | both editions agree and the Ming 南陽堂 recension prints the line a third time character for character, which collates it and adds no witness to it; what modern software carries under this name is the 月系 star placed off the month — two quantities wearing one name, and the block is what stops that being an inference, placing 天德 and 月德 off the 流年 three lines above it |
| the 四化 at 庚 | sihua: quanshu | 5 | 《全書》's 庚日武陰同為首 | the two editions of this book disagree about the line, and the reading kept is the one the references compute; a 中州派 manual sides with it, which is a school agreeing and not a witness. The Ming 南陽堂 recension, read on the plate, prints a third form — 庚日武同相為者, 科 天同 and 忌 天相 — so the line has three readings and the shipped one is the only one two voices agree on; 中國絕學 第一冊, read on the plate, works the same reading in a chart of its own and heads a section on the two 化科 with it, which is a second school agreeing and still not a witness |
| the 四化 at 壬, the reading two schools move | sihua: zuofu | 2 | the 中州派 manual's 「壬梁紫輔武宿是」, and 《紫微斗數 北派》 p. 四—一八二, which draws 左輔 with 科 in a worked example decidable on that reading alone | nothing that runs: two modern manuals agreeing on one cell, against a reading both editions of 《全書》 and the Ming 南陽堂 recension carry |
| the five phases the board is inked by | — | 2 | three passages of 卷二, which give a phase to every star it places bar thirteen | both editions agree on all three word for word, which collates the text and leaves the doctrine one work's, and 輔弼二星屬土 is the second edition's alone and followed as a lone reading. **The rung is contested by the rule that says so** — three passages of one juan is rung 4, not rung 2 — and it was left standing rather than moved in the commit that wrote the rule |
| 身主 | — | 5 | the 身主 line of 卷二, which reads 「子午人火玲星」 in both editions | the collation confirms there is nothing to resolve it with: the defect is the work's and not a copyist's, and the induction first made here was withdrawn |
Bā Zì 八字
| Quantità | Scuola | Rung | Si regge su | Verificata su |
|---|---|---|---|---|
| the 大運 — the day the first decade opens | luckGranularity: shichen | 1 | three days of the calendar to a year lived, counted to the 節 and stopped at the whole double hour | lunar-javascript, inside the 479 charts of the row above; the `minute` reading is the same rule carried further and parts from it by up to ten days |
| the count of the five elements | — | 0 | the element of each stem and each branch, arithmetic over the eight characters, zeroes included | the basis is the tier-1 material the seasonal states and the 納音 already share; the count adds no rule |
lìzhù 曆注
| Quantità | Scuola | Rung | Si regge su | Verificata su |
|---|---|---|---|---|
| 建除十二直 — the officer of the day | — | 1 | the count anchored on the 節, doubling the day a term opens on | lunar-javascript, every day from 2000 to 2039: officer and day pillar together 14600/14600 |
| 二十八宿值日 | — | 1 | one unbroken count of days, crossing a 節 where 建除 doubles | lunar-javascript 14600/14600 with the 七政 of each lodge, on an epoch over-determined by the seven-day lock the names carry |
| 十二神 — the 黃黑道 | — | 4 | 《協紀辨方書》卷七, which rejects the two accounts it was handed and derives its own: 天罡 laid on the 建 | the source works three months out in full and all three fall out of the rule |
| 年神 — twenty-six bearings | shensha: xieji | 5 | 《協紀辨方書》卷三, each one whose position the source states outright and completely in its own entry | nothing: one text, quoted. What the source says about each beyond the bearing is 宜忌 and does not travel |
| 四德 — 天德, 月德 and their 合 | — | 4 | 《協紀辨方書》卷四, enumerated whole | each 德 is confirmed by its own 五合 partner, which is the source stating one fact twice; against lunar-javascript over 3650 days, 天德合 100 %, 月德合 99.86 %, 天德 99.59 %, 月德 99.29 %, the residue unexplained and left standing |
| 神煞 — the seven a day carries or does not | shensha: xieji | 5 | 《協紀辨方書》卷五, each keyed to the month's branch, the season, or the day branch alone | nothing beyond the source's own enumeration. 《御定星歷考原》 was got as the second witness and cannot serve: the copy held is 卷一 of six, which carries no 神煞. Its 六合 derivation is confirmed on the plate, fuller than the 協紀's citation of it |
niánmìng 年命
| Quantità | Scuola | Rung | Si regge su | Verificata su |
|---|---|---|---|---|
| 本命 and 行年 | count: sui | 5 | 《遁甲演義》卷一 遁甲錯誤須檢點, which states the rule and states leaving it out as a defect | nothing to run it against. The 四庫全書提要 singles the doctrine out as one other books do not carry, which is a witness to its rarity and not a second statement of it |
I programmi su cui è stato verificato
Implementazioni indipendenti, ciascuna lavoro di qualcun altro, eseguite sugli intervalli che le righe qui sopra dichiarano. I testi citati accanto non hanno un collegamento, perché dare un indirizzo a un testo significherebbe affermare quale edizione si sta citando.