分类归档：历法与方志考据 / Calendrics & Chronology

最后修订时间：2026年10月18日

Chujia Kuang (邝楚嘉)

Independent Researcher, Jieyang, Guangdong, China

Corresponding author: Chujia Kuang, Independent Researcher, Jieyang, Guangdong, China. E-mail: chirca@163.com

Abstract

In 1978, at Leigudun in Suixian, Hubei (now Suizhou), an archaeological team opened a Warring States tomb. Its occupant was Yi, a ruler of the state of Zeng whom later ages call Marquis Yi of Zeng. Beside its famous set of chime bells, the tomb yielded a lacquered wooden clothes chest, catalogued E.66. On its lid, around a large seal-script character for the Dipper, are written twenty-eight star names, with a dragon drawn to the east and a tiger to the west. This note uses that object as its starting point and sets out the two schemes by which the ancients divided the sky. The first is the twenty-eight mansions (二十八宿), star regions laid out one to each night of the moon’s eastward circuit, along the ecliptic and near the celestial equator; the note explains what a mansion is, lists the twenty-eight names in their fixed order in four groups of seven, and records the four guardian animals — Azure Dragon, Dark Warrior, White Tiger and Vermilion Bird — showing that their left-and-right arrangement follows from the single posture of observing while facing south. The second scheme is the three enclosures (三垣): the Purple Forbidden, Supreme Palace and Heavenly Market Enclosures around the north celestial pole. The note then explains how the grid was used, at dusk and at dawn, to record position and to infer the season. The note is descriptive; it contains no predictive claims.

Keywords: twenty-eight mansions; three enclosures; four images; celestial equator; ecliptic; determinative star; Zeng Hou Yi lacquer chest; Chinese astronomy

1. The character su comes first

In 1978, at Leigudun in Suixian, Hubei — now Suizhou — an archaeological team opened a tomb of the Warring States period.

The occupant’s name was Yi, a ruler of the state of Zeng whom later ages call Marquis Yi of Zeng. The tomb is best known for its set of chime bells. From the same tomb was also recovered a lacquered wooden clothes chest, catalogued E.66.

According to the inscription in the chest, it was originally called hu (匫) and was used for storing clothes. The lid, taken off, is covered with a layer of black lacquer, on which a diagram is drawn in vermilion.

At the centre of the diagram is a large character 斗 (dou, the Dipper) in seal script. Written around it in a circuit are twenty-eight star names. To the east a dragon is drawn, to the west a tiger.

This is the earliest physical object bearing the twenty-eight mansions that can be seen today. The burial of Marquis Yi of Zeng is generally dated to around 433 BCE. That is to say, as early as the fifth century BCE these twenty-eight names were already common knowledge, casually painted on the lid of an everyday chest.

Where those twenty-eight names came from, and what use the division served — this must begin with the character 宿 (su).

Su (宿) is the su of lodging for the night.

The moon moves through the starry sky, shifting eastward by about thirteen degrees a day; each day it passes into a different region of the sky. One region per night, like a traveller changing inns each day. Following this rhythm, the ancients divided the stars near the ecliptic and the celestial equator in turn into twenty-eight star regions, each region called a mansion (宿, su).

The twenty-eight mansions are also called the twenty-eight lodgings or the twenty-eight stars.

Wang Chong of the Eastern Han has a sentence in the Lunheng that puts the matter plainly:

二十八宿为日月舍，犹地有邮亭，为长吏廨矣。
(The twenty-eight mansions are the lodgings of the sun and moon, just as on the earth there are postal stations, which are the public offices of the senior officials.)

The mansions in the sky are the relay stations of the sun and moon, comparable to the postal stations on earth, the official buildings where passing officers rest.

This sentence names the use of the twenty-eight mansions: recording position. Where the sun, moon and the five planets have reached, the ancients recorded with a phrase such as “entering such-and-such a mansion by so many degrees”. Within each mansion a star was chosen as a marker, called the determinative star (距星, juxing), and positions were measured from it. Spread the twenty-eight cells across the sky, and the sky acquires a scale.

The earlier notes dealt with scales on the ground: how the year comes about, how the month, how the day is cut, how the double-hour is divided. The twenty-eight mansions are the set in the sky. Only the two sets together make a complete base map.

In the “Spring Offices” section of the Zhouli there is an office called the Fengxiang shi, whose charge was precisely the positions of the year, the month, the nodes, the day and the twenty-eight stars. The number appears in the classics very early.

2. The twenty-eight names

The twenty-eight mansions have a fixed order, beginning with the Horn and running from west to east, in the direction of the moon’s motion. Each of the four directions has seven mansions, and each direction has a name:

Direction	Name	The seven mansions
East	Azure Dragon (Canglong)	Horn · Neck · Root · Room · Heart · Tail · Winnowing Basket
North	Dark Warrior	Dipper · Ox · Girl · Emptiness · Rooftop · Encampment · Wall
West	White Tiger	Legs · Bond · Stomach · Hairy Head · Net · Turtle Beak · Three Stars
South	Vermilion Bird	Well · Ghost · Willow · Star · Extended Net · Wings · Chariot

Several of these characters are awkward to read, so their sounds are noted first: 亢 is read kàng, 氐 dī, 箕 jī, 娄 lóu, 昴 mǎo, 觜 zī, 参 shēn, 轸 zhěn.

There are two “Dipper” characters, easily confused. The seven stars of the Northern Dipper are in the northern sky, the ladle that scoops wine; the Dipper mansion is in the southern sky with only six stars, called the Six Stars of the Southern Dipper. The two “Dippers” face each other across the celestial pole, and poets often set the two against each other.

Among the twenty-eight mansions, four can still be recognised today by looking up:

The core of the Heart mansion is called Antares (心宿二); the old books call it Great Fire (大火, dahuo); it is reddish in colour and one of the brightest stars in the whole sky.

The Hairy Head mansion is a small knot of stars; a person with good eyesight can make out six or seven crowded together. This is the Pleiades cluster, which the West calls the Seven Sisters.

The Three Stars mansion corresponds to today’s Orion; its three central stars are strung in a horizontal line and are easiest to recognise on a winter night.

The Dipper mansion is the Six Stars of the Southern Dipper just mentioned.

3. Four animal names, from one posture

The seven eastern mansions are the Azure Dragon, the seven western the White Tiger, the seven southern the Vermilion Bird, the seven northern the Dark Warrior. The four animals divide the twenty-eight mansions into four parts, together called the four images (四象). The Liji, “Qu li, part one”, has a line:

行前朱鸟而后玄武，左青龙而右白虎。
(In marching, the Vermilion Bird in front and the Dark Warrior behind, the Azure Dragon on the left and the White Tiger on the right.)

Kong Yingda of the Tang, subcommenting on it, says plainly:

朱鸟、玄武、青龙、白虎，四方宿名也。
(The Vermilion Bird, the Dark Warrior, the Azure Dragon and the White Tiger are the names of the mansions of the four directions.)

The names of the mansions of the four directions. What these four names denote is the four directional divisions of the twenty-eight mansions, not the four animals themselves. By the way, the Liji writes the Vermilion Bird as 朱鸟 (zhuniao); later ages mostly write it 朱雀 (zhuque); the same group is meant.

Then where do the left and right in “Azure Dragon on the left, White Tiger on the right” come from?

From a single action: standing facing south.

The ancients were accustomed to watching the stars facing south. At dusk around the spring equinox, standing facing south, overhead is the southern sky, where the Vermilion Bird is; on the left hand is the eastern sky, where the Azure Dragon has just risen; on the right hand is the western sky, where the White Tiger is half hidden below the horizon; behind is the northern sky, where the Dark Warrior has already set. The four directions and the four animals are all derived from the single posture of “observing while facing south”. Change the direction you stand facing, and left and right are thrown into confusion.

Some scholars doubt that so neat a pairing of the four images with the twenty-eight mansions was a later addition, made after the Han. But since the dragon, the tiger and the twenty-eight mansions share a single lid on the lacquer chest, it does not hold that the pairing had no basis at all.

4. There are also three “cities” in the sky

The twenty-eight mansions govern a belt of sky along the ecliptic and the celestial equator. The region near the north celestial pole has another scheme of division, called the three enclosures (三垣, san yuan).

Yuan (垣) means a wall. The three enclosures are three walled cities.

The Purple Forbidden Enclosure rings the north celestial pole and was taken by the ancients as the dwelling of the celestial emperor; the pole star and the Dipper are both within it. The Supreme Palace Enclosure lies south of the Purple Forbidden, the imperial court, its star names mostly official titles. The Heavenly Market Enclosure lies further south, a marketplace, its star names including a crowd of “rows of shops”, “rows of carts”, the Dipper and the grain measure, all implements of trade and of measuring.

The three enclosures together with the twenty-eight mansions are, in the eyes of the ancients, the whole of the starry sky. The scheme took its roughly final form in the Sui and Tang, and can be traced further back. The Forbidden City in Beijing was laid out with reference to the Purple Forbidden Enclosure, holding to the principle of “setting out the capital in the image of heaven”.

The preceding note, on time, mentioned that the earth’s axis of rotation slowly traces a circle in space. The position of the pole star shifts, and the “pole star” is not forever fixed on one particular star. That matter is left to the note on precession. Here one sentence need only be kept in mind: the Kaogong ji, speaking of fixing direction at night, uses the five characters 夜考之极星, “at night examine the pole star”. That pole star is in the Purple Forbidden Enclosure.

5. How the grid is used

The twenty-eight mansions are not there for show. The ancients used them chiefly to watch four moments:

A mansion just rising in the east at dusk is called “dusk rising”; a mansion hanging due south at dusk is called “dusk culmination”; rising in the east before dawn is “dawn rising”; hanging due south before dawn is “dawn culmination”.

The four moments are four rulers. A given mansion reaches due south at dusk only during a certain stretch of the year. Record them day by day, and the season at hand can be inferred backwards.

Set this aspect aside for now; the note devoted to the solar terms will take up “which mansion is due south, and what farm work should be done”.

One further aspect of the twenty-eight mansions is to be used later: direction.

How is one to tell east, west, north and south at night? All the stars in the sky are turning; only one place barely moves. Find that place, and direction is had.

Closing

Return to that lacquer chest.

The twenty-eight characters on the chest lid can still be made out one by one by a person today. In the two thousand four hundred years since, the positions of the stars have shifted a good deal, yet the names and the order of this set of cells have scarcely changed.

The ancients drew a grid on the sky in order to know where they stood. Look up, and a certain mansion is directly overhead; direction is had.

What follows in this series is that matter. In the years and months before the compass, how a pole, two shadows and one unmoving star measured out east, west, north and south.

A&Q

What are the twenty-eight mansions?

The twenty-eight star regions divided out along the ecliptic and near the celestial equator, each region called a mansion. Wang Chong of the Eastern Han, in the Lunheng, says “the twenty-eight mansions are the lodgings of the sun and moon” — the relay stations of sun and moon.

How are the twenty-eight mansions arranged?

From the Horn, running from west to east, in the direction of the moon’s motion. Seven in each of the four directions, matched with the Azure Dragon, the Dark Warrior, the White Tiger and the Vermilion Bird.

In “Azure Dragon on the left, White Tiger on the right”, from which side are left and right reckoned?

Standing facing south. Facing south, the left hand is the eastern sky, the right hand the western sky, and behind is the northern sky.

What are the three enclosures?

The three “cities” near the north celestial pole — the Purple Forbidden Enclosure, the Supreme Palace Enclosure and the Heavenly Market Enclosure — arranged from north to south. Together with the twenty-eight mansions they make the whole of the starry sky as the ancients saw it.

What use are the twenty-eight mansions?

Recording position, and also reading the seasons. Watch a mansion’s dusk rising, dusk culmination, dawn rising and dawn culmination, and the season at hand can be inferred backwards.

Note by Chujia Kuang

The string of names of the twenty-eight mansions looks at first like a heap of unrecognisable characters. Read smoothly, one sees that it is in fact a table — the sky divided into twenty-eight cells, so that the positions of sun, moon and the five planets have something to be stated against. Looking up at the sky, what the ancients first understood was this: everything in the sky is moving, but it moves by rule, and can be recorded and inferred. The calendar grew out of that rule. This is groundwork; later, in treating the solar terms and precession, we shall return to this table.

About the author

Chujia Kuang (邝楚嘉), pen name Jiayan Yide (嘉言一得), is an independent researcher in Jieyang, Guangdong, devoted to archiving verifiable primary sources and calendar data.

This column writes about calendrics, solar terms, the stems and branches and orientation, and also about what he reads. The pen name 嘉言一得 comes from the 史记: 愚者千虑，必有一得, “the foolish man’s thousand thoughts must yield one gain”. Where something is written wrongly, corrections are welcome.

ORCID: 0009-0002-7650-833X ｜ OpenAlex: A5151908354

Note on the text

This is a popular exposition, and it makes no claim to original research. The astronomical constants cited are standard values. This English version is issued alongside the Chinese original as the same article in another language. The author’s name is 邝楚嘉, Chujia Kuang; he writes under the pen-name 嘉言一得 (Jiayan Yide).

Declarations

Funding. This research received no external funding.

Competing interests. The author declares no competing interests.

Data availability. No new data were created or analysed in this study.

Language. The original Chinese text is deposited as a companion file to this record (古人的星空：三垣与二十八宿).

References

[1] Zhouli [周礼][M]. Pre-Qin period.

[2] Kaogong ji [考工记][M]. Pre-Qin period.

[3] Liji [礼记][M]. Western Han.

[4] Wang Chong. Lunheng [论衡][M]. Eastern Han.

Images in this article are illustration only; they are not evidence and not a basis for any statement of fact.

This article is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
You are free to reprint, redistribute, or build upon it, including for commercial purposes, provided you credit the original author as 「嘉言一得（Chujia Kuang 邝楚嘉）」 and include a link to the original publication.
Copyright in the excavated Qin and Han bamboo-slip photographs and in the transmitted texts quoted here remains with their respective rights holders and is not covered by this licence.
For the full terms of the licence, see: https://creativecommons.org
Related articles

• Three Calendars, Three Kinds of Time — the three scales and what each is set to.
• Ten Stems, Twelve Branches: Why Only Sixty Pairings Remain — why ten stems and twelve branches give sixty, not a hundred and twenty.
• Lichun Is Not a Day: The Solar Term Is an Instant — a solar term is an instant, not a day.
• The Gnomon and the Sun at Its Utmost — shadow-measurement before the Dengfeng observatory.
• The Twenty-Eight Mansions on the Lacquer Box from the Tomb of Marquis Yi of Zeng — a lower bound for the four images.

Chinese editions: 古人的星空：三垣与二十八宿（the original Chinese text of this essay）

Site hubs: Glossary ｜ Calendar Reform ｜ Publications

All data in this article were collected and compiled by Chujia Kuang.

---

作者：嘉言一得（邝楚嘉）｜独立研究者，致力于可核验的原始文献与历法数据归档。
Author: Chujia Kuang (邝楚嘉), pen name Jiayan Yide (嘉言一得) — independent researcher devoted to archiving verifiable primary sources and calendar data.
平实记录 · 数据集 · 预印本：GitHub / WordPress / Zenodo（DOI: 10.5281/zenodo.22788686 ／ 10.5281/zenodo.23028692）
Records · Datasets · Preprints: GitHub / WordPress / Zenodo
学术身份 · Academic identity：ORCID 0009-0002-7650-833X ｜ OpenAlex A5151908354

最后修订时间：2026年10月18日

Chujia Kuang (邝楚嘉)

Independent Researcher, Jieyang, Guangdong, China

Corresponding author: Chujia Kuang, Independent Researcher, Jieyang, Guangdong, China. E-mail: chirca@163.com

Abstract

This note sets out the three great circles of the sky: the ecliptic (黄道), the celestial equator (天赤道) and the lunar path (白道). The ecliptic is the sun’s annual path, the circle traced on the celestial sphere by the plane of the earth’s orbit; the Hanshu, “Treatise on Astronomy”, also calls it the middle road and the light road. The celestial equator is the earth’s equator extended outward, and it serves as the base circle of measurement. The two circles cross at twenty-three and a half degrees, 23°26′ today, and meet at two points; the days on which the sun reaches them are the spring and autumn equinoxes. The angle was measured, in 103 CE with a bronze ecliptic instrument made to imperial order. The moon’s path is inclined to the ecliptic by a little over five degrees, 5°09′ on average and swinging between 4°57′ and 5°19′. Solar and lunar eclipses hang on this: they occur only when new moon or full moon falls near the node where the two paths cross — within about eighteen degrees for a solar eclipse and twelve for a lunar one. The nodes themselves regress westward, completing a circuit in eighteen and a half years; this was computed by Liu Hong in the second century. The note closes with the Hanshu‘s “nine roads of the moon” and Shen Kuo’s remark that all such names are forced names, divisions made for calculation and not roads that exist in the sky. The note is descriptive; it contains no predictive claims.

Keywords: ecliptic; celestial equator; lunar path; equinoxes; eclipse node; nine roads of the moon; Shen Kuo

“The sun comes up in the east and goes down in the west.”

That sentence describes the layer of rotation. The earth turns once, and the sun draws a circle around us in the sky, east in the morning and west in the evening.

The sun has another road besides. It travels slowly, taking a whole year to complete one circuit, and it does not travel by day.

By day the sun is bright and the stars beside it cannot be seen. But when the sky has not yet gone fully dark after sunset, or when it is just greying before dawn, one can sometimes see the sun and a few bright stars at the same time. Compared day by day, one finds that the sun’s position among the stars is always shifting, about one degree eastward each day.

In a year it shifts through a full circuit and returns to its old place.

This road is called the ecliptic (黄道).

1. The ecliptic

The Hanshu, “Treatise on Astronomy”, has a line:

「日有中道，月有九行。中道者，黄道，一曰光道。」
(“The sun has a middle road, the moon has nine courses. The middle road is the ecliptic, also called the light road.”)

The road the sun travels is called the middle road, also the light road, also the ecliptic. Three names written side by side — which shows that in the Han it was not yet a designation that could not be otherwise.

Middle road and light road are both plainer than “ecliptic”: the road that lies in the middle, the road that gives light.

Present-day astronomy describes it as a geometric object — the plane of the earth’s orbit extended into the sky, intersecting the celestial sphere in a great circle. The sun’s position through the year falls on this circle.

The Hanshu has another line further on: 「日之所行为中道，月、五星皆随之也。」(“Where the sun travels is the middle road; the moon and the five planets all follow it.”) The road the sun travels is the main road, and the moon and the five planets all travel along it.

2. The celestial equator

There is another road in the sky, whose name is “equator” with “celestial” before it.

It is the plane of the earth’s equator extended out in both directions, cutting the celestial sphere in a great circle. Where the equator is beneath your feet, the celestial equator is the circle directly overhead. Its two ends are the north and south celestial poles.

Zhang Heng wrote of it in the Hunyi:

「赤道横带浑天之腹。」
(“The equator lies as a band across the belly of the celestial sphere.”)

A band across the belly of the sky, like a belt drawn horizontally around the sphere’s waist. This circle divides the celestial sphere through its exact middle.

This line is less busy than the ecliptic, but its use is the most solid: it is the base circle of measurement. The stars in the sky are all turning, and the axis of that turning is the celestial pole; to say where a star is, one must first say how far it lies from the celestial equator and how many degrees from the pole, and only then can different places be speaking of the same thing.

3. The two roads are twenty-three and a half degrees apart

The ecliptic and the celestial equator do not coincide. They cross at a slant, enclosing twenty-three and a half degrees, written today as 23°26′.

Zhang Heng wrote of this angle too:

「黄道斜带其腹，出赤道表里各二十四度。」
(“The ecliptic lies aslant across its belly, emerging on either side of the equator by twenty-four degrees each way.”)

The twenty-four degrees are ancient degrees. The ancients divided the full circuit of the sky into a little more than three hundred and sixty-five degrees, not three hundred and sixty; converted, these twenty-four degrees come to just over twenty-three degrees today.

This angle was measured. In the fifteenth year of Yongyuan of the Eastern Han, that is 103 CE, the court ordered the making of a bronze ecliptic instrument (黄道铜仪) for the Grand Astrologer, built expressly for measuring the ecliptic. When Zhang Heng wrote the Hunyi, only twenty or thirty years had passed since the instrument was made; the figures in the Hunyi were read off from it, mark by mark.

The two circles cross, meeting at two points. The two days on which the sun reaches those points are the spring and autumn equinoxes. Zhang Heng has another line that puts it more directly: 「黄道斜截赤道者，即春、秋分之去极也。」(“The point where the ecliptic cuts the equator aslant is precisely the distance from the pole at the spring and autumn equinoxes.”)

Because of this angle, the sun in the course of a year runs north of the equator for a while and south of it for a while. The Hanshu, “Treatise on Astronomy”, describes this road: north it reaches 东井 and lies close to the pole; south it reaches 牵牛 and lies far from the pole.

Running north, it shines more directly and for longer; running south, the reverse. The origin of cold and heat lies in these twenty-three and a half degrees.

4. The lunar path

The moon too has a road of its own, called the lunar path (白道).

The moon circles the earth; the plane of its orbit projected onto the sky is also a circle. This circle does not coincide with the ecliptic, differing by a little over five degrees — the figure today is 5°09′ on average, and it is always moving back and forth between 4°57′ and 5°19′.

Five degrees does not sound like much, but the solar and lunar eclipses of the sky all hang on it.

At new moon the moon is between the sun and the earth; at full moon the earth is between. If the two roads were to lie one on the other, there should be a solar eclipse on the first of every month and a lunar eclipse on the fifteenth. But they do not coincide: on most firsts the moon slips past above or below the sun; on most fifteenths it slips past above or below the earth’s shadow.

Only when the new or full moon falls precisely near the node of the two roads is there an eclipse. This “near” has a figure: about eighteen degrees for a solar eclipse and about twelve for a lunar one.

There is one more thing. Those two nodes are themselves restless, forever moving westward, by a little over nineteen degrees a year, completing one circuit in eighteen and a half years. This was computed by Liu Hong of the Eastern Han, who also applied it to the reckoning of lunar eclipses. That was the second century CE.

5. Nine roads, and in fact only one

The Hanshu, “Treatise on Astronomy”, continues under “the moon has nine courses”:

「月有九行者：黑道二，出黄道北；赤道二，出黄道南；白道二，出黄道西；青道二，出黄道东。」
(“The moon has nine courses: two black roads, emerging north of the ecliptic; two red roads, emerging south of it; two white roads, emerging west of it; two blue roads, emerging east of it.”)

Black, red, white and blue, two each, eight in all; add the ecliptic itself, and they make nine. This is the “nine courses of the moon”.

So the white road spoken of by the people of the Han was the two roads west of the ecliptic, not the moon’s whole orbit. The white road spoken of earlier is a usage that became current only later. The two white roads are not the same thing.

Shen Kuo of the Northern Song put the matter most clearly. In the Mengxi bitan he says:

「历法，天有黄、赤二道，月有九道。此皆强名而已，非实有也。」
(“In calendar-making, the sky has the two roads, ecliptic and equator, and the moon has nine courses. These are all forced names and nothing more; they do not really exist.”)

Forced names are names assigned by force. The sky has an ecliptic and an equator, and the moon’s road has nine courses; these names were all fastened on, and were not there in the sky to begin with.

He goes on to explain how the names arose:

「日之所由，谓之黄道；南北极之中，度最均处，谓之赤道。月行黄道之南，谓之朱道；行黄道之北，谓之黑道。黄道之东，谓之青道；黄道之西，谓之白道。黄道内外各四，并黄道为九。」
(“Where the sun travels is called the ecliptic; the point midway between the north and south poles, where the degrees are most even, is called the equator. The moon moving south of the ecliptic is the vermilion road; moving north of it, the black road. East of the ecliptic is the blue road; west of it, the white road. Inside and outside the ecliptic, four each, and together with the ecliptic, nine.”)

The road the sun travels is called the ecliptic; the place exactly midway between the two poles, where the degrees are most even, is called the equator; the moon running south of the ecliptic is the vermilion road, north of it the black road, and east and west one each.

Then comes the weighty line:

「故因其合散，分为数段，每段以一色名之，欲以别算位而已。如算法用赤筹、黑筹，以别正负之数。历家不知其意，遂以谓实有九道，甚可嗤也。」
(“So, because of their joining and parting, the path is divided into several segments, and each segment is named with a colour, only in order to distinguish the computed positions. It is like using red counters and black counters in arithmetic to distinguish positive and negative numbers. Calendar-makers who do not understand this have taken it to mean that nine roads really exist; this is quite laughable.”)

Dividing the road into segments and giving each a colour is only to make the computed positions easy to tell apart, just as in arithmetic red counters and black counters are used to record positive and negative numbers. Are there really nine roads? He found the idea laughable.

6. Three roads, each with its own charge

Looking back, these three roads are three rulers the ancients set up in the sky.

Road	Whose it is	How far it differs
Ecliptic	the sun’s	23°26′ from the celestial equator
Lunar path	the moon’s	about 5°09′ from the ecliptic
Celestial equator	carries no one, serves only as the base circle of measurement	—

Whether the sun on this road runs north or south decides whether the ground is a season of growth or of harvest. At what angle the moon on the lunar path stands to the sun decides whether this night is new moon or full moon, and whether there is an eclipse. The celestial equator gives off no light of its own and carries no one, but the positions on the other two roads must all be measured against it as their base circle.

Having gone round the circle, back to names.

None of these three roads was carved in the sky. The sun, the moon and the earth each travel their own way, and none meant to leave a mark for anyone. People below looked on, and looking for long enough traced the paths they had taken into circles, and gave them names.

The earlier piece on time said that the graduations are added by human beings. The roads in the sky are the same.

Shen Kuo’s “quite laughable” laughs at taking an algorithm for the sky. Nine roads, ecliptic and equator — all are segments that people divided out in order to compute positions. To divide them conveniently is a skill; to suppose, once divided, that the sky really has these nine roads, is a mistake.

The sun still rises in the east today and still sets in the west at night. It has shifted another degree or so along the ecliptic, and no one has seen it.

What comes next is the grid of the sky: the three enclosures and the twenty-eight mansions.

A&Q

What is the ecliptic?

The road the sun travels through the sky in the course of a year. It is the great circle formed where the plane of the earth’s orbit, extended into the sky, cuts the celestial sphere; the Hanshu, “Treatise on Astronomy”, calls it the middle road, and also the light road.

How far apart are the ecliptic and the equator?

They enclose 23°26′. Zhang Heng’s Hunyi writes it as 「出赤道表里各二十四度」 (“emerging on either side of the equator by twenty-four degrees each way”), which is in ancient degrees.

What is the lunar path?

The road the moon travels, differing from the ecliptic by 5°09′ on average, and swinging back and forth between 4°57′ and 5°19′.

Why is there not an eclipse every month?

The two roads do not coincide. There is an eclipse only when the new or full moon falls precisely near the node; about eighteen degrees for a solar eclipse and twelve for a lunar one.

Are there really nine roads in the moon’s course?

No. They are segment names used in computation. Shen Kuo says in the Mengxi bitan, 「此皆强名而已，非实有也」 (“these are all forced names and nothing more; they do not really exist”).

Note by Chujia Kuang

This piece is about three lines. The names sound abstruse, but their origin is thoroughly concrete: all of them were measured out by eye. The three terms — ecliptic, lunar path, celestial equator — appear very frequently in the later almanacs, yet what they actually mean is seldom explained. That is set out clearly here, so that when we come to the starry sky and to orientation there will be somewhere to stand. The passage from Shen Kuo I have copied at some length, because it puts the matter plainly: an algorithm is an algorithm, and the sky is the sky.

About the author

Chujia Kuang (邝楚嘉), pen name Jiayan Yide (嘉言一得), is an independent researcher in Jieyang, Guangdong, devoted to archiving verifiable primary sources and calendar data.

This column writes about calendrics, solar terms, the stems and branches and orientation, and also about what he reads. The pen name 嘉言一得 comes from the 史记: 愚者千虑，必有一得, “the foolish man’s thousand thoughts must yield one gain”. Where something is written wrongly, corrections are welcome.

ORCID: 0009-0002-7650-833X ｜ OpenAlex: A5151908354

Note on the text

This is a popular exposition, and it makes no claim to original research. The astronomical constants cited are standard values. This English version is issued alongside the Chinese original as the same article in another language. The author’s name is 邝楚嘉, Chujia Kuang; he writes under the pen-name 嘉言一得 (Jiayan Yide).

Declarations

Funding. This research received no external funding.

Competing interests. The author declares no competing interests.

Data availability. No new data were created or analysed in this study.

Language. The original Chinese text is deposited as a companion file to this record (天上三条道：黄道、白道、赤道).

References

[1] Ban Gu. Hanshu [汉书][M]. Eastern Han.

[2] Zhang Heng. Hunyi [浑仪][M]. Eastern Han.

[3] Shen Kuo. Mengxi bitan [梦溪笔谈][M]. Song dynasty.

Images in this article are illustration only; they are not evidence and not a basis for any statement of fact.

This article is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
You are free to reprint, redistribute, or build upon it, including for commercial purposes, provided you credit the original author as 「嘉言一得（Chujia Kuang 邝楚嘉）」 and include a link to the original publication.
Copyright in the excavated Qin and Han bamboo-slip photographs and in the transmitted texts quoted here remains with their respective rights holders and is not covered by this licence.
For the full terms of the licence, see: https://creativecommons.org
Related articles

• Three Calendars, Three Kinds of Time — the three scales and what each is set to.
• Ten Stems, Twelve Branches: Why Only Sixty Pairings Remain — why ten stems and twelve branches give sixty, not a hundred and twenty.
• Lichun Is Not a Day: The Solar Term Is an Instant — a solar term is an instant, not a day.
• The Gnomon and the Sun at Its Utmost — shadow-measurement before the Dengfeng observatory.
• The Twenty-Eight Mansions on the Lacquer Box from the Tomb of Marquis Yi of Zeng — a lower bound for the four images.

Chinese editions: 天上三条道：黄道、白道、赤道（the original Chinese text of this essay）

Site hubs: Glossary ｜ Calendar Reform ｜ Publications

All data in this article were collected and compiled by Chujia Kuang.

---

作者：嘉言一得（邝楚嘉）｜独立研究者，致力于可核验的原始文献与历法数据归档。
Author: Chujia Kuang (邝楚嘉), pen name Jiayan Yide (嘉言一得) — independent researcher devoted to archiving verifiable primary sources and calendar data.
平实记录 · 数据集 · 预印本：GitHub / WordPress / Zenodo（DOI: 10.5281/zenodo.22788686 ／ 10.5281/zenodo.23028692）
Records · Datasets · Preprints: GitHub / WordPress / Zenodo
学术身份 · Academic identity：ORCID 0009-0002-7650-833X ｜ OpenAlex A5151908354

最后修订时间：2026年10月18日

Chujia Kuang (邝楚嘉)

Independent Researcher, Jieyang, Guangdong, China

Corresponding author: Chujia Kuang, Independent Researcher, Jieyang, Guangdong, China. E-mail: chirca@163.com

Abstract

A day is said to have twenty-four hours, and a watch keeps to that figure. The figure is not exact. One revolution of the earth, measured against the fixed stars, takes 23 hours 56 minutes 4 seconds; the missing four minutes are consumed by the earth’s motion along its orbit, so that the sun returns overhead only after the earth has turned a little further. Both statements are true; they count different things. This note asks where time comes from, and answers that it was never produced: it is the reading a person records against several periodic motions. The second is defined by the vibration of a caesium-133 atom, because only a motion that repeats without end can serve as a ruler. Han texts already say as much: the Shizi, the Zhuangzi and the Mozi tie rest to time as firmly as motion. Three rulers — day, month and year — are then set against their motions; all three give non-integer readings, and the whole difficulty of the calendar grows from those remainders. The same motion, measured against the sun, the moon or the fixed stars, yields different figures: 365.2422 solar days to a tropical year, but 366.2422 sidereal days. The remainders are shown to have been counted on the ground, in the 1461-day cycle and the four-minute nightly shift of the stars. Finally, because the earth’s rotation is slowing under tidal friction, the leap second is needed to hold civil time in step. The note is descriptive; it contains no predictive claims.

Keywords: time; periodic motion; sidereal day; solar day; tropical year; precession; leap second

“A day has twenty-four hours.”

Everyone says it, and the watch runs by it.

But it is not exact.

One full revolution of the earth — from this instant, turning back to the direction it faced before — takes 23 hours 56 minutes 4 seconds. Nearly four minutes short of twenty-four hours.

The four minutes are not lost. Another motion eats them: the earth turns, and at the same time it travels forward. By the time it has completed one turn, the sun is no longer in its old position, and the earth must turn a little further before the sun comes back overhead.

A little further each day, and there are the four minutes.

So “a day is twenty-four hours” and “one turn of the earth is 23 hours 56 minutes” are both correct; they are simply not counting the same thing.

One step further down: where does time come from?

1. Only periodic motion can serve as a ruler

We are used to thinking of time as a kind of thing, like the water of a river, flowing of itself, with the clock merely measuring how much has passed.

In its modern definition it is not that at all. The hardest example is the second.

In 1967 the Thirteenth General Conference on Weights and Measures redefined the second:

The second is the duration of 9 192 631 770 periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the caesium-133 atom.

That definition is still in use. The 2018 revision of the International System of Units left the second built on this frequency, only written differently.

There is a detail here worth turning over: why choose the vibration of an atom, of all things, to define how long a second is?

Because only a motion that repeats, over and over, can serve as a ruler.

The moon circling the earth, the earth circling the sun, the earth turning on its own axis, the electron in an atom jumping back and forth between two levels — these are all periodic: after an interval they begin again. The day one of them stops repeating, it stops being usable as a ruler.

So “time” is not a thing that was discovered. It is the reading a person records against several periodic motions.

Which motions, then?

2. Two thousand years ago the Chinese were already saying this

Start with 宇 and 宙.

The people of the pre-Qin period joined space and time in a single word, 宇宙. The word is still in use today, but its original sense was far more substantial than the present one:

「天地四方曰宇，往古来今曰宙。」——《尸子》
(“Heaven, earth and the four directions are called yu; what is past and what is to come are called zhou.” — Shizi)

宇 is heaven, earth and the four directions — that is space; 宙 is from antiquity to the present — that is time.

The Huainanzi, “Qisu xun”, puts it almost identically: 往古来今谓之宙，四方上下谓之宇 (“what runs from antiquity to the present is called zhou; the four directions, above and below, are called yu”).

The Zhuangzi, “Gengsang Chu”, takes another angle:

「有实而无乎处者，宇也；有长而无本剽者，宙也。」
(“That which has reality but no fixed place is yu; that which has length but no beginning or end is zhou.”)

Space is solid and real, yet no fixed place can be found for it; time keeps growing, yet no head and no tail can be found for it.

(One caution: the original Shizi was long since lost, and the present text is a reconstruction by later hands; this line is cited in the commentary to the Shishuo xinyu, “Paitiao”, so in quoting it one must say so.)

The Mohists went a step further.

In the Mozi, “Jingshang”, there are two pairs of lines worth reading together:

「久，弥异时也。」「宇，弥异所也。」
「动，或徙也。」「止，以久也。」
(“Jiu [duration] is what covers the different times; yu is what covers the different places.” “Motion is the shifting of a region; rest is measured by jiu [duration].”)

The first pair says: 久, that is time, strings every different moment together; 宇 strings every different place together.

The second is more important. 或 is the ancient form of 域, so 「动，或徙也」 means: motion is the migration of a region of place. And 「止，以久也」 means: rest, too, must be measured by time (久).

People of more than two thousand years ago already saw it clearly: without time, not even “rest” can be stated. Motion and rest both hang on time at either end.

This is the same direction as present-day physics; only the ancients said it in five characters.

Take one piece of positive evidence. The Shangshu, “Yao dian”, records Yao’s command:

「钦若昊天，历象日月星辰，敬授民时。」
(“Reverently follow Heaven above; observe and compute the sun, the moon and the stars; respectfully bestow the seasons on the people.”)

历象 is observing and computing; 敬授民时 is bestowing the computed seasons on the common people. The seasons do not drop from the sky; they must be read out of the motion of the sun, the moon and the stars, and then passed on. This sentence is, by and large, the governing principle of every calendar that followed.

3. Three rulers, all of them motion

The oldest units of time — not one of them was “made” by anyone; all were “counted”.

Unit	Which motion is counted	Reading
Day	One rotation of the earth (counted against the sun)	1 mean solar day = 24 hours
Month	One cycle of the moon’s phases (counted against the phase)	1 synodic month ≈ 29.53 days
Year	One circuit of the earth around the sun (counted against the vernal equinox)	1 tropical year ≈ 365.2422 days

Three motions, three units. The motion comes first, the unit after; it is not that a unit came first and was then fitted to a motion.

But the three rulers share one defect: none of them counts out to a whole number.

A year is not exactly 365 days. A month is not exactly 29 and a half days. A day is not exactly 24 hours.

What the calendar has had to deal with from its first day is these remainders. The lichun, the solar terms and zhengyue treated in the earlier pieces all have their root here.

4. The same motion, a different reference, a different count

This is the most important part of the whole piece.

To count a motion one must first find something “not moving” to serve as reference. But a different reference gives a different count. The same motion can be reported as several different times.

Unit	Counted against the sun or the phase	Counted against the fixed stars	Difference
Day	mean solar day = 24 hours	mean sidereal day = 23 hours 56 minutes 4 seconds	about 3 minutes 56 seconds
Month	synodic month = 29.53059 days	sidereal month = 27.32166 days	about 2.21 days
Year	tropical year = 365.24220 days	sidereal year = 365.25636 days	about 20 minutes 24 seconds

(For the last digits of the differences, each source has its own slight variation. Take the difference between the solar day and the sidereal day: some write 3 minutes 56 seconds, others 3 minutes 57 seconds.)

Why, counted against the sun, is the day longer? The reason in one sentence:

The earth turns, and at the same time travels forward.

By the time it has completed a full turn and returned to its former direction, the sun has moved forward along the orbit by about 1 degree. The earth must turn that much more before the sun comes back directly overhead.

So the solar day = one rotation + the little while needed to turn about 1 degree more.

Push this out over a whole year and a counter-intuitive number appears:

In one tropical year there are 365.2422 solar days, but 366.2422 sidereal days.

That is to say, in the course of this year the earth in fact turned more than 366 times; yet we lived through only a little more than 365 days.

The extra turn is not a miscalculation. It is eaten by the question “what is taken as the reference”: one day eats 3 minutes 56 seconds, and over 365 days that comes to exactly one whole turn.

This layer also settles the earlier question of “why there are three calendars”. None of the three calendars miscalculated; each simply took a different base: take the sun as base and you get the rhythm of the “year”; take the fixed stars and you get the rhythm of the “day”; take the phase and you get the rhythm of the “month”.

5. Those remainders were counted on the ground

The remainders were counted out by the ancients, day by day.

Take the year. Through long observation the ancients found that from the moment the sun’s shadow returns to its former place to the moment it coincides with the previous one, the interval is 1461 days, exactly four years. Divide by four, and a year is 365.25 days.

Then what of the extra day? 1461 = 365 × 4 + 1; that surplus day is the leap day.

Why the leap day comes every four years — the answer is here, not in a formula.

Now the 3 minutes 56 seconds. A little each day, accumulated over 365 days, makes exactly one whole day, and that is the extra 366th turn. The account balances.

There is one more thing visible merely by looking up: the same star rises about four minutes earlier each night than the night before.

Four minutes a day is nothing to notice. A month makes two hours; over a year, the whole sky of stars has turned over once. That the star patterns of the four seasons change has its root in these 3 minutes 56 seconds.

6. Motion changes, and time changes with it

If time really is the reading of motion, there is a layer further down still: motion itself changes, and time should change with it.

This has already been measured in astronomy.

The earth’s rotation is slowly slowing. The main cause is tidal friction. The moon’s gravity raises the sea water, and the sun does its part as well, a smaller one; as the tide grinds across the ocean floor it carries friction, which is to say it puts a brake on the earth’s rotation.

How much slower? About every hundred years, a day grows by nearly two milliseconds. (Values differ slightly from source to source, roughly between 1.7 and 2 milliseconds; this is an order of magnitude, not a figure fixed to the decimal place.)

Two milliseconds is imperceptible to a person today. Put it into geological time, though, and it is startling: by one study, a day in the age of the dinosaurs was only about 23.5 hours.

Hence the leap second.

The length of the second kept by the atomic clock is constant; the earth’s rotation is not. The two run each on its own, and in time they separate. From 1972 the world agreed: when the gap grows large, a second is added to Coordinated Universal Time, holding the difference within 0.9 of a second.

From 1972 to the present, twenty-seven such seconds have been added, every one of them an addition, never once a subtraction. All of them “additions” shows precisely that the earth is slowing.

In 2022 the Twenty-seventh General Conference on Weights and Measures passed a resolution: no later than 2035, leap seconds will no longer be introduced.

There is a still slower layer: the direction of the motion itself is also drifting.

The earth’s axis of rotation traces a cone in space, and the vernal equinox retreats westward year by year. The direction in which a given star lay two thousand years ago is no longer where it lies today. This is called precession. It is too slow; the earlier pieces have left it alone, and a piece of its own comes later.

Closing

So back to the question at the start: how is time produced?

It was never produced. It is the reading a person records against several motions.

One turn of the earth is a day; one cycle of the moon’s phases is a month; one circuit of the earth around the sun is a year.

Motion first, graduations after. Motion was already there; the graduations are added by human beings.

The ancients recorded these readings, aligned them, cut them into segments, and so had the calendar. And the neat twenty-four hours on our watches today is in fact a ruler that has been smoothed — the true sun does not run evenly; it is the “mean sun” that smooths it on its behalf. The earlier piece on clock time and true solar time spoke of “one uneven, one even”; this is what it meant. The sidereal day and the solar day here are another pair of differences within the same family.

What comes next is the several roads in the sky — the one the sun travels, the one the moon travels, and the one used to measure time.

Note by Chujia Kuang

The reason for taking up this subject is to bring the word “time” down to the ground. It is not a thing that can be accumulated; it is the reading a person records against several periodic motions. Once this layer is understood, the earlier pieces have their origin: why the start of spring is no more than a single instant, why the first month shifted several times, why the solar terms differ by a few hours from year to year — all those remainders grew from here.

Note on the text

This is a popular exposition, and it makes no claim to original research. The astronomical constants cited are standard values. This English version is issued alongside the Chinese original as the same article in another language. The author’s name is 邝楚嘉, Chujia Kuang; he writes under the pen-name 嘉言一得 (Jiayan Yide).

Declarations

Funding. This research received no external funding.

Competing interests. The author declares no competing interests.

Data availability. No new data were created or analysed in this study.

Language. The original Chinese text is deposited as a companion file to this record (时间是怎么产生的？).

References

[1] Shi Jiao. Shizi [尸子][M]. Pre-Qin period.

[2] Zhuang Zhou. Zhuangzi [庄子][M]. Pre-Qin period.

[3] Mo Di. Mozi [墨子][M]. Pre-Qin period.

[4] Shangshu [尚书][M]. Pre-Qin period.

[5] Liu An. Huainanzi [淮南子][M]. Western Han.

[6] Liu Yiqing. Shishuo xinyu [世说新语][M]. Northern and Southern Dynasties.

Images in this article are illustration only; they are not evidence and not a basis for any statement of fact.

This article is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
You are free to reprint, redistribute, or build upon it, including for commercial purposes, provided you credit the original author as 「嘉言一得（Chujia Kuang 邝楚嘉）」 and include a link to the original publication.
Copyright in the excavated Qin and Han bamboo-slip photographs and in the transmitted texts quoted here remains with their respective rights holders and is not covered by this licence.
For the full terms of the licence, see: https://creativecommons.org
Related articles

• Three Calendars, Three Kinds of Time — the three scales and what each is set to.
• Ten Stems, Twelve Branches: Why Only Sixty Pairings Remain — why ten stems and twelve branches give sixty, not a hundred and twenty.
• Lichun Is Not a Day: The Solar Term Is an Instant — a solar term is an instant, not a day.
• The Gnomon and the Sun at Its Utmost — shadow-measurement before the Dengfeng observatory.
• The Twenty-Eight Mansions on the Lacquer Box from the Tomb of Marquis Yi of Zeng — a lower bound for the four images.

Chinese editions: 时间是怎么产生的？（the original Chinese text of this essay）

Site hubs: Glossary ｜ Calendar Reform ｜ Publications

All data in this article were collected and compiled by Chujia Kuang.

---

作者：嘉言一得（邝楚嘉）｜独立研究者，致力于可核验的原始文献与历法数据归档。
Author: Chujia Kuang (邝楚嘉), pen name Jiayan Yide (嘉言一得) — independent researcher devoted to archiving verifiable primary sources and calendar data.
平实记录 · 数据集 · 预印本：GitHub / WordPress / Zenodo（DOI: 10.5281/zenodo.22788686 ／ 10.5281/zenodo.23028692）
Records · Datasets · Preprints: GitHub / WordPress / Zenodo
学术身份 · Academic identity：ORCID 0009-0002-7650-833X ｜ OpenAlex A5151908354

最后修订时间：2026年10月18日

Chujia Kuang (邝楚嘉)

Independent Researcher, Jieyang, Guangdong, China

Corresponding author: Chujia Kuang, Independent Researcher, Jieyang, Guangdong, China. E-mail: chirca@163.com

Abstract

A birth certificate records clock time. The double-hour of birth is something else. This note works through the difference between true solar time and mean solar time, and the three corrections that lie between a recorded clock time and the double-hour it must be read against. The first is daylight saving time: China observed it for six years, 1986 to 1991, and a time recorded in that window must be put back an hour. The second is longitude: Beijing time is the mean solar time of 120°E, not the local time of Beijing, and since China spans some sixty-two degrees of longitude under a single standard line, the correction reaches more than two hours in the west; the rule is one degree to four minutes, applied as 4 minutes × (local longitude − 120°). The third is the equation of time, at most about sixteen minutes fast and fourteen minutes slow, which matters only when a birth time sits close to a double-hour boundary; two of its causes are the elliptical orbit and the obliquity of the ecliptic, and it passes through zero four times a year. The note then sets out the long-standing dispute over whether the 子 hour, which spans 23:00 to 01:00 and two calendar days, should turn the day at its initial (子初) or at its midpoint (子正), citing the Gexiang xinshu and the Sanming tonghui, and declines to choose between them. It closes with a four-step checklist: establish the place of birth, check daylight saving, correct for longitude, and apply the equation of time only on a boundary. The note is descriptive; it contains no predictive claims.

Keywords: true solar time; mean solar time; Beijing time; longitude correction; daylight saving time; equation of time; double-hour; 子 hour

1. Two rulers

True solar time: taken from the sun’s actual position. The moment the sun reaches the true south and rises to its highest is the local noon. This is what the sundial measures.

Mean solar time: an imaginary second sun, moving uniformly along the celestial equator, taking exactly 24 hours every day of the 365-day year. This is what the clock keeps.

Why imagine one? Because the real one does not move uniformly.

The earth’s orbit is an ellipse; by Kepler’s laws it runs fast through perihelion in early January and slow through aphelion in early July. The sun’s speed across the sky changes throughout the year, and a clock regulated by it would make the days unliveable.

So mean solar time is uniform and true solar time is not. Two rulers, measuring different things.

The ten fifty-six at Chengdu in the opening is a figure after the longitude correction, with the equation of time not yet applied. The equation of time is at most some tens of minutes; the fifth section of this note deals with it — when it must be reckoned and when it may be left aside.

2. Beijing time is not the time of Beijing

This is the layer most easily confused.

Beijing time is the mean solar time of the meridian 120°E — not the local time of Beijing. Beijing actually lies at about 116.4°E, so when the clock points to twelve, Beijing is still about 14 minutes short of noon.

More to the point: our country spans some 62 degrees of longitude from east to west (73°E to 135°E), and uses the single line of 120°E for the whole territory.

The standard time of a time zone is only an artificial time that everyone agrees to keep in common; it is not the local time of every place within that zone.

History has tried otherwise. In the Republican period the country was at one time divided into five time zones:

Time zone	Central meridian
Central Plains	120°E
Longshu	105°E
Xin-Zang	90°E
Kunlun	82.5°E (some say 75°)
Changbai	127.5°E (some say 135°)

Later they were unified into one. Unification has its advantages; the price is that clock time and local solar time fall out of step.

3. One degree, four minutes

The earth turns 360 degrees in 24 hours, so 1 degree = 4 minutes.

The formula is simple:

真太阳时 = 北京时间 + 4 分 ×（当地经度 − 120°）+ 均时差
(True solar time = Beijing time + 4 minutes × (local longitude − 120°) + the equation of time.)

Longitude corrections for the main cities:

City	Longitude	Correction
Ürümqi	87.6°	−130 min
Lhasa	91.1°	−116 min
Kunming	102.7°	−69 min
Chengdu	104.1°	−64 min
Xi’an	108.9°	−44 min
Guangzhou	113.3°	−27 min
Shenzhen	114.1°	−24 min
Beijing	116.4°	−14 min
Shanghai	121.5°	+6 min
Harbin	126.6°	+26 min

Two examples:

Shenzhen, born at 9:10 Beijing time. Correction −24 min → true solar time 8:46. 9:10 is the 巳 hour, 8:46 is the 辰 hour. One double-hour apart.

Shanghai, born at 23:55 Beijing time. Correction +6 min → true solar time 0:01 the next day. It crosses zero o’clock, and the day-pillar changes.

4. The cut of daylight saving

There is a further error that is artificial, and the most easily overlooked.

From 1986 to 1991 the whole country observed daylight saving for six years: from 2 a.m. on the first Sunday in mid-April the clock was put forward one hour, and from 2 a.m. on the first Sunday in mid-September it was put back.

• 4 May — 14 September 1986 (the first year was special)
• 12 April — 13 September 1987
• 10 April — 11 September 1988
• 16 April — 17 September 1989
• 15 April — 16 September 1990
• 14 April — 15 September 1991

From 5 April 1992 it was suspended.

For those born in these six years, the time recorded on the birth certificate is the time after the clock was put forward, and one hour must be subtracted to restore it. Many people do not know they were once “put forward”, and cast their charts straight away, wrong at the first step.

5. The equation of time: this layer can be left to the last

The difference obtained by subtracting mean solar time from true solar time is the equation of time.

• Over the year: at most about 16 minutes fast and about 14 minutes slow
• It is zero four times in a year
• Two causes: the elliptical orbit (about ±7.66 min) and the obliquity of the ecliptic (about ±9.87 min)

Set the three levels of error side by side, and their weights are very far apart:

Level	Source of error	Maximum	Must it be reckoned?
First	daylight saving time	±60 min	Yes
Second	longitude	up to more than 2 hours	Yes
Third	equation of time	±16 min	only when it sits on a double-hour boundary

That is: for most people the longitude correction is enough. The equation of time needs to be brought out only when the birth time lies exactly on the boundary between double-hours, and within twenty minutes of it either way.

Do not give the impression that one “must reckon to the second” — that is bluffing.

6. The 子 hour: how those two hours are reckoned

The 子 hour runs 23:00–01:00 and spans two days. Here there is an old question of a thousand years.

School	Position	Basis
Day changes at 子初	23:00 already counts as the new day	the Yuanhai ziping, the method of “raising the double-hour from the day”; folk tradition; “the double-hour may not be divided”
Day changes at 子正	00:00 is the boundary. For a midnight 子 hour the day-pillar takes the present day and the hour-pillar takes the 子 hour	the Sanming tonghui, “On the Double-Hours”; the Xin Tangshu, “Treatise on the Calendar”, “古历分日，起于子半”

The clearest statement in the sources is Zhao Youqin of the Yuan (courtesy name Yuandu), in the Gexiang xinshu:

若子时则上半时在夜半前属昨日，下半时在夜半后属今日。
(In the case of the 子 hour, the first half falls before midnight and belongs to the previous day; the second half falls after midnight and belongs to the present day.)

Wan Minying of the Ming, in the Sanming tonghui, “On the Double-Hours”, has a passage identical to it word for word:

子时则上半时在夜半前，属昨日；下半时在夜半后，属今日。
(The 子 hour: the first half falls before midnight and belongs to the previous day; the second half falls after midnight and belongs to the present day.)

Each view has its own textual lineage; this note makes no recommendation — take whichever your own transmission prescribes.

Hence the folk saying, “神仙难断子时命 — not even the immortals can settle the fate of a 子-hour birth”.

But more important than the dispute is the order:

先扣夏令时 → 再按经度换成真太阳时 → 看是否落进子时区间 → 最后才谈早晚子时
(First deduct daylight saving → then convert by longitude to true solar time → see whether it falls within the 子-hour interval → only then speak of the early or late 子 hour.)

Without the longitude correction first, whether it is even the 子 hour is not yet settled, and arguing about the early or late 子 hour is meaningless.

7. A four-step checklist

From birth time to the hour-pillar, there are just these four steps:

1. Establish the place of birth — not “where are you from” but “at which hospital”. A province’s difference may be a double-hour’s difference.
2. Check daylight saving — for anyone born in 1986–1991, subtract one hour first.
3. Apply the longitude correction — formula: Beijing time + 4 min × (local longitude − 120°).
4. Apply the equation of time only on a boundary — if after the first two steps the time is still more than half an hour from the boundary, it may be ignored.

Closing

Back to the title: the clock time of birth is not the double-hour of birth.

Where is the difference? One comes from the sun, one from the clock, and one is a hand pulling the clock forward — and what casting a chart needs is the one from the sun.

The clock is unified across the country; the sun is not.

When the double-hours were discussed earlier, it was said that the ancients measured them with a sundial — measured, that is, the sun. So we are not inventing any new rule, but recovering the same ruler.

What comes next is the stems and branches — the ten heavenly stems and the twelve earthly branches: what those twenty-two signs actually are.

Note on the text

This is a popular exposition, and it makes no claim to original research. The astronomical constants cited are standard values. This English version is issued alongside the Chinese original as the same article in another language. The author’s name is 邝楚嘉, Chujia Kuang; he writes under the pen-name 嘉言一得 (Jiayan Yide).

Declarations

Funding. This research received no external funding.

Competing interests. The author declares no competing interests.

Data availability. No new data were created or analysed in this study.

Language. The original Chinese text is deposited as a companion file to this record (钟表上的出生时间，不等于你出生的时辰).

References

[1] Xu Dasheng. Yuanhai ziping [渊海子平][M]. Song dynasty.

[2] Ouyang Xiu, Song Qi. Xin Tangshu [新唐书][M]. Song dynasty.

[3] Zhao Youqin. Gexiang xinshu [革象新书][M]. Yuan dynasty.

[4] Wan Minying. Sanming tonghui [三命通会][M]. Ming dynasty.

Images in this article are illustration only; they are not evidence and not a basis for any statement of fact.

This article is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
You are free to reprint, redistribute, or build upon it, including for commercial purposes, provided you credit the original author as 「嘉言一得（Chujia Kuang 邝楚嘉）」 and include a link to the original publication.
Copyright in the excavated Qin and Han bamboo-slip photographs and in the transmitted texts quoted here remains with their respective rights holders and is not covered by this licence.
For the full terms of the licence, see: https://creativecommons.org
Related articles

• Three Calendars, Three Kinds of Time — the three scales and what each is set to.
• Ten Stems, Twelve Branches: Why Only Sixty Pairings Remain — why ten stems and twelve branches give sixty, not a hundred and twenty.
• Lichun Is Not a Day: The Solar Term Is an Instant — a solar term is an instant, not a day.
• The Gnomon and the Sun at Its Utmost — shadow-measurement before the Dengfeng observatory.
• The Twenty-Eight Mansions on the Lacquer Box from the Tomb of Marquis Yi of Zeng — a lower bound for the four images.

Chinese editions: 钟表上的出生时间，不等于你出生的时辰（the original Chinese text of this essay）

Site hubs: Glossary ｜ Calendar Reform ｜ Publications

All data in this article were collected and compiled by Chujia Kuang.

---

作者：嘉言一得（邝楚嘉）｜独立研究者，致力于可核验的原始文献与历法数据归档。
Author: Chujia Kuang (邝楚嘉), pen name Jiayan Yide (嘉言一得) — independent researcher devoted to archiving verifiable primary sources and calendar data.
平实记录 · 数据集 · 预印本：GitHub / WordPress / Zenodo（DOI: 10.5281/zenodo.22788686 ／ 10.5281/zenodo.23028692）
Records · Datasets · Preprints: GitHub / WordPress / Zenodo
学术身份 · Academic identity：ORCID 0009-0002-7650-833X ｜ OpenAlex A5151908354

最后修订时间：2026年10月18日

Chujia Kuang (邝楚嘉)

Independent Researcher, Jieyang, Guangdong, China

Corresponding author: Chujia Kuang, Independent Researcher, Jieyang, Guangdong, China. E-mail: chirca@163.com

Abstract

The day, for most people, begins at daybreak or at zero o’clock. For the ancients it began at eleven at night. This note sets out why: the double-hour 子 (zi), the first of the twelve, is the moment of the day farthest from the sun, the hour of greatest yin — and it is exactly there that the day is held to turn, just as the winter solstice, the day of the year farthest from the sun, is the point at which the single yang returns. The twelve double-hours are then presented as a set of names drawn not from an abstract scale but from the ordinary round of the day; two of the twelve are mealtimes, which is itself evidence that the ancients ate twice a day. A double-hour is exactly two hours, and after the Tang each was divided into an initial (初, chu) and a medial (正, zheng) half, giving twenty-four divisions whose length matches the modern hour — the origin of the word xiaoshi, “small hour”. The note traces the twelvefold division against the rival schemes of two, five and ten divisions, and dates its final form to the Taichu calendar reform of 104 BCE; it then gives the length of the ke (刻) under the hundred-ke and ninety-six-ke systems (14.4 minutes before 1670, 15 minutes after) and the five watches (更, geng) of the night. It closes by noting that the double-hour was measured by the sundial, and that clock time and solar time are not the same thing. The note is descriptive; it contains no predictive claims.

Keywords: double-hour; 子 hour; midnight; twelve double-hours; ke; five watches; sundial; solar time

1. Why the starting point is at midnight

What time does a day begin?

Most people would answer: at daybreak. Or at zero o’clock.

The ancients’ answer is stranger — eleven at night.

This was not chosen at random. Behind it lies a whole understanding of time; once this point is grasped, the twelve double-hours that follow will stand.

First, a counter-intuitive fact: of the twelve double-hours, the first, 子 (zi), is the moment of the day farthest from the sun.

The 子 hour runs from 23:00 to 01:00; the sun has moved to the far side of the earth, directly beneath our feet. It is the time of day when the yin is at its height.

Yet the ancients set the beginning of the day precisely at this darkest moment.

Why?

The Sanming tonghui, in its chapter “On the Double-Hours”, gives an explanation of remarkable aptness:

若子时，则上半时在夜半前，属昨日。下半时在夜半后，属今日。亦犹冬至得十一月中气，一阳来复，为天道之初耳。
(In the case of the 子 hour, the first half falls before midnight and belongs to the previous day; the second half falls after midnight and belongs to the present day. It is just as the winter solstice, which takes the second qi of the eleventh month, when the single yang returns, is the beginning of the way of heaven.)

The meaning is: the 子 hour stands to a day as the winter solstice stands to a year.

The winter solstice is the day of the year with the shortest daylight and the greatest yin, but it is also the turning point at which the yang begins to rise again — the saying “at the winter solstice the single yang is born”. The 子 hour is the same: it is the darkest moment of the day, and it is the moment at which the yang begins to return to life.

So the Leijing tuyi says: “子者阳生之初” — “the 子 is the beginning of the birth of the yang.”

The beginning of the day lies not at the brightest point but at the instant of turning, at the darkest point. This way of thinking is of one heart with the saying “when adversity reaches its extreme, good fortune comes”.

Two corroborations:

• Li Chunfeng of the Tang records in the Xin Tangshu, “Treatise on the Calendar”: “古历分日，起于子半” — “the old calendars divided the day, beginning at the middle of the 子 [hour]” — taking the midpoint of the 子 hour, that is, zero o’clock, as the boundary of the day.
• Fu Sheng of the early Han, in the Shangshu dazhuan: “周以十一月为正，色尚赤，以夜半为朔” — “the Zhou took the eleventh month as the corrected beginning, favoured the colour red, and took midnight as the first of the month” — the Zhou took midnight as the beginning of a month.

Thus there are three “beginnings”, each answering to the next:

The beginning of the year is zhengyue; the beginning of a month is the first day; the beginning of a day is the middle of the 子 hour.

A widely circulated statement should be corrected in passing: “the 子 hour is twelve o’clock at midnight” is not accurate.

Twelve o’clock is called 子正 (zizheng), the midpoint of the 子 hour; the 子 hour itself is 23:00 to 01:00, a full two hours.

By the same token, the 午 hour is 11:00 to 13:00, and it is 午正 (wuzheng) that is twelve o’clock.

On a chart this distinction makes a real difference: a birth at half past eleven at night and a birth at half past midnight are separated by only one hour, yet fall respectively in 子初 and 子正.

2. The twelve double-hours are twelve pictures of daily life

The ancients did not first have “子, 丑, 寅, 卯” and then fit them to time. They first had the daily round, and then gave each stretch a name.

Double-hour	Clock	The picture within the name
子	23—1	Midnight (夜半, yeban). 姑苏城外寒山寺，夜半钟声到客船 — “Outside Gusu city, at Cold Mountain Temple, the midnight bell reaches the traveller’s boat”
丑	1—3	Cock-crow (鸡鸣, jiming). 鸡声茅店月，人迹板桥霜 — “The cock’s cry, the moon over a thatched inn; footprints, frost on a plank bridge”
寅	3—5	Dawn (平旦, pingdan). 夜与日交替之际，《孟子》说的「平旦之气」 — “the hour when night and day alternate; the ‘dawn air’ spoken of in the Mencius“
卯	5—7	Sunrise (日出, richu). 日出而作，日入而息 — “out at sunrise and in at sunset”
辰	7—9	Meal time (食时, shishi). 吃早饭的时候 — the time of the morning meal
巳	9—11	Forenoon (隅中, yuzhong). 临近中午 — drawing near to noon
午	11—13	Midday (日中, rizhong). 太阳升到中天 — the sun rises to mid-heaven
未	13—15	The sun’s decline (日昳, ridie). 日头开始偏西 — the sun begins to lean west
申	15—17	Late-afternoon meal (晡时, bushi). 吃第二顿饭的时候 — the time of the second meal
酉	17—19	Sunset (日入, riru). 山中相送罢，日暮掩柴扉 — “the farewells in the hills are done; at dusk the brushwood gate is closed”
戌	19—21	Dusk (黄昏, huanghun). 月上柳梢头，人约黄昏后 — “the moon rises above the willow tops; a tryst is kept after dusk”
亥	21—23	People at rest (人定, rending). 奄奄黄昏后，寂寂人定初 — “after the fading dusk, at the first still hour when people settle”

See it now — of the twelve names, two are mealtimes: 辰 is “meal time”, the morning meal, and 申 is the “late-afternoon meal”, the second meal.

This shows that the ancients ate only twice a day. One meal between seven and nine in the morning, one between three and five in the afternoon, with nothing in between.

This detail is more useful than memorising the twelve earthly branches. It reminds you that the double-hours did not drop from the sky as an abstract scale; they grew out of daily life.

3. A double-hour is exactly two hours

The twelve double-hours divide a day and night evenly into twelve parts, so one double-hour equals two hours — a day of exactly twenty-four hours.

This “twenty-four” is no accident. From the Tang onward each double-hour was divided into an initial (初) and a medial (正) half of one hour each, so that the twelve double-hours became twenty-four parts. This is China’s own “twenty-four hours to the day”, and each part so produced is exactly equal in length to today’s hour.

More interesting still is the origin of the word “small hour” (xiaoshi, 小时).

Qian Daxin of the Qing, in the Shijiazhai yangxin lu, “The Twenty-Four Hours”, puts it clearly:

一日分十二时，每时又分为二，曰「初」，曰「正」，是为二十四小时。
(A day is divided into twelve double-hours, and each double-hour is again divided into two, called the initial and the medial, giving the twenty-four hours.)

The ancients called a double-hour a great hour (大时, dashi). Later, to keep step with the internationally current twenty-four-hour system, each great hour was split in two, and the half so produced was called the small hour (小时, xiaoshi).

Half of a great hour is a small hour.

We say the word every day, and few have wondered why it carries the character “small” — small by contrast with the “great hour” that is the double-hour.

4. Why twelve parts and not ten

How many parts to cut the day into — the ancients tried more than one scheme.

The crudest division is into two: sunrise to sunset is “day”, sunset to sunrise is “night”, called the “two-part system of dark and light”. This does not even amount to a unit, since the length of day and night changes with the seasons.

The Chunqiu Zuozhuan, under the fifth year of Duke Zhao, has a further line: “日之数十，故有十时” — “the numbers of the sun are ten, and so there were ten hours.” This shows that there was once a ten-double-hour system, dividing the day into ten parts.

There was later also a five-double-hour system.

What won in the end was twelve. The reasons are not hard to see:

• A year has twelve lunar months, and twelve was already the period the ancients knew best.
• The year-star (Jupiter) takes about twelve years to circuit the sky, and the ancients used it to record the years.
• The twelve earthly branches form a set of their own, and can be made to mark direction, month and double-hour alike.

The correspondence of double-hour to direction is equally direct: 子 is true north, 午 is true south.

The sun rises to its highest point in the true south at the 午 hour (upper culmination), and sinks to its lowest in the true north at the 子 hour (lower culmination). So “midday” is 午 and “midnight” is 子 — the names of the double-hours correspond directly to the sun’s position in the sky.

5. When it was settled

• The Zhouli already has the phrase “十有二辰” — “the twelve 辰”; the shape is visible in the Shang and Zhou.
• In 104 BCE, after Emperor Wu of Han promulgated the Taichu calendar, it took substantially its final form.
• Zhao Yi of the Qing, in the Gaiyu congkao: “In the Taichu period of the Han the Taichu calendar was put into effect; ‘其以一日分十二时，而以干支为纪’ — ‘it divided a day into twelve double-hours and recorded them by the stems and branches’.”
• The pairing of the twelve names (midnight, cock-crow, …) with the twelve earthly branch names was completed only about the beginning of the first century CE.

So the statement “the twelve double-hours have existed since antiquity” is not accurate. The names and the branches were matched only two thousand years ago.

6. How long, exactly, is a ke

The ancients measured time with a water-clock (a bronze clepsydra), dividing a day and night into one hundred ke (刻); the system is thought to have existed already in the Shang.

A day and night is 1,440 minutes ÷ 100 ke = 14.4 minutes to the ke, that is, 14 minutes 24 seconds.

The trouble is that one hundred ke cannot be divided evenly by twelve: each double-hour comes to 8⅓ ke, which is very inconvenient to reckon.

In the ninth year of Kangxi of the Qing (1670) the hundred ke were changed to ninety-six ke, giving “twelve double-hours to the day, eight ke to the double-hour, fifteen minutes to the ke, sixty seconds to the minute” — each ke exactly 15 minutes, in step with the internationally current hours, minutes and seconds.

So is a “ke” 14.4 minutes or 15 minutes? Both are right; it depends on the dynasty: 14.4 before Kangxi, 15 after.

In passing, a widely circulated statement should be laid aside —

“午时三刻” is not twelve o’clock noon.

By the hundred-ke system: the initial of the 午 hour begins at 11:00, and three ke = 3 × 14.4 minutes ≈ 43.2 minutes.

“午时三刻” is about 11:43.

When a novel says “execution at 午时三刻”, it means around twenty to twelve in the morning, not noon. The notion that this is “when the yang is at its height” likewise derives from this double-hour.

7. Why “the third watch” and “酉正三刻” differ so much

At night there was another clock: the night was divided into five watches (更, geng), each announced by a drum, and each watch again into five points (点, dian).

Watch	Equivalent today
First watch	19:12
Second watch	21:36
Third watch	0:00
Fourth watch	2:24
Fifth watch	4:48

One night is forty ke (daytime being sixty ke), each watch about 144 minutes, each point about 28.8 minutes.

Here there is a dividing line of period: before the Song the watch-and-point system was used; after the Song the double-hour-and-ke system.

So when the Shuihu zhuan says “at the third watch” and the Honglou meng says “酉正三刻”, this is not the author writing at random; it is the difference of period between two systems of reckoning time.

8. One last thing easily overlooked

These double-hours were measured by the ancients with a sundial (日晷).

The sundial reads the sun’s actual position: the moment the sun reaches its highest point in the true south is the local “medial of the 午 hour”.

The twelve o’clock on our clocks today is another matter altogether.

The same two characters “午时” point, for the ancients, to one moment, and for today’s clock to another — and across most of China the two do not agree; in some places they are two hours apart.

This is for later.

Closing

The beginning of the day is at midnight, not at daybreak.

Once this is grasped, the twelve double-hours are no longer twelve cold graduations — where the sun goes, that is what people should be doing; this scale grew out of the soil.

What comes next is zhengyue — which was not originally the first month.

Note on the text

This is a popular exposition, and it makes no claim to original research. The astronomical constants cited are standard values. This English version is issued alongside the Chinese original as the same article in another language. The author’s name is 邝楚嘉, Chujia Kuang; he writes under the pen-name 嘉言一得 (Jiayan Yide).

Declarations

Funding. This research received no external funding.

Competing interests. The author declares no competing interests.

Data availability. No new data were created or analysed in this study.

Language. The original Chinese text is deposited as a companion file to this record (古人的一天，从晚上十一点开始).

References

[1] Meng Ke. Mengzi [孟子][M]. Pre-Qin period.

[2] Zuozhuan [左传][M]. Pre-Qin period.

[3] Zhouli [周礼][M]. Pre-Qin period.

[4] Fu Sheng. Shangshu dazhuan [尚书大传][M]. Western Han.

[5] Ouyang Xiu, Song Qi. Xin Tangshu [新唐书][M]. Song dynasty.

[6] Wan Minying. Sanming tonghui [三命通会][M]. Ming dynasty.

[7] Zhang Jiebin. Leijing tuyi [类经图翼][M]. Ming dynasty.

[8] Shi Nai’an. Shuihu zhuan [水浒传][M]. Ming dynasty.

[9] Qian Daxin. Shijiazhai yangxin lu [十驾斋养新录][M]. Qing dynasty.

[10] Zhao Yi. Gaiyu congkao [陔余丛考][M]. Qing dynasty.

[11] Cao Xueqin. Honglou meng [红楼梦][M]. Qing dynasty.

Images in this article are illustration only; they are not evidence and not a basis for any statement of fact.

This article is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
You are free to reprint, redistribute, or build upon it, including for commercial purposes, provided you credit the original author as 「嘉言一得（Chujia Kuang 邝楚嘉）」 and include a link to the original publication.
Copyright in the excavated Qin and Han bamboo-slip photographs and in the transmitted texts quoted here remains with their respective rights holders and is not covered by this licence.
For the full terms of the licence, see: https://creativecommons.org
Related articles

• Three Calendars, Three Kinds of Time — the three scales and what each is set to.
• Ten Stems, Twelve Branches: Why Only Sixty Pairings Remain — why ten stems and twelve branches give sixty, not a hundred and twenty.
• Lichun Is Not a Day: The Solar Term Is an Instant — a solar term is an instant, not a day.
• The Gnomon and the Sun at Its Utmost — shadow-measurement before the Dengfeng observatory.
• The Twenty-Eight Mansions on the Lacquer Box from the Tomb of Marquis Yi of Zeng — a lower bound for the four images.

Chinese editions: 古人的一天，从晚上十一点开始（the original Chinese text of this essay）

Site hubs: Glossary ｜ Calendar Reform ｜ Publications

All data in this article were collected and compiled by Chujia Kuang.

---

作者：嘉言一得（邝楚嘉）｜独立研究者，致力于可核验的原始文献与历法数据归档。
Author: Chujia Kuang (邝楚嘉), pen name Jiayan Yide (嘉言一得) — independent researcher devoted to archiving verifiable primary sources and calendar data.
平实记录 · 数据集 · 预印本：GitHub / WordPress / Zenodo（DOI: 10.5281/zenodo.22788686 ／ 10.5281/zenodo.23028692）
Records · Datasets · Preprints: GitHub / WordPress / Zenodo
学术身份 · Academic identity：ORCID 0009-0002-7650-833X ｜ OpenAlex A5151908354

最后修订时间：2026年10月18日

Chujia Kuang (邝楚嘉)

Independent Researcher, Jieyang, Guangdong, China

Corresponding author: Chujia Kuang, Independent Researcher, Jieyang, Guangdong, China. E-mail: chirca@163.com

Abstract

Zhengyue (正月) is ordinarily taken to mean the first month of the year. The two were not always the same thing. This note sets out the san zheng (三正), the three reckonings of the year’s beginning that the sources attribute to the Xia, Shang and Zhou, and lists the month each took as its head. The Qin is added to that list: its year opened in the tenth month, and the Qin did not call that month zhengyue — it renamed zhengyue as duanyue (端月) instead, while the four seasons continued to run by the Xia reckoning. The restoration is traced to the Taichu li of 104 BCE, which fixed the head of the year at the first month; that arrangement has held for two thousand years, with four interruptions after it (Wang Mang, the Wei emperor Ming, and twice under the Tang, including Wu Zetian’s manufacture of an auspicious date in 698). The note then separates three things the sources keep distinct — the head of the year, the name zhengyue, and the pairing of seasons with months — citing Sima Zhen’s commentary on the Shiji, and records Wang Li’s doubt as to whether the san zheng can be credited at all. It closes with Shen Kuo’s unadopted proposal for a purely solar calendar of months divided by solar terms. The note is descriptive; it contains no predictive claims.

Keywords: zhengyue; san zheng; head of the year; Taichu calendar; Zhuanxu calendar; Qin calendar; Shen Kuo

1. The three zheng: which month heads the year

Start with the claim itself.

The two characters zhengyue (正月) did not originally mean the first month.

They meant the month in which the head of the year falls — the first month of the year. And which month that was has been changed many times in the course of more than two thousand years.

Travel back to the Qin and celebrate the New Year there, and that month is the tenth.

The ancients fixed the months by the seven stars of the Dipper. Whichever direction the handle points to establishes (建, jian) that month; the months 寅, 卯 and 辰 are named in this way.

Which month stood at the head of the year differed from dynasty to dynasty:

Dynasty	Dipper handle	Head of the year, by today’s reckoning
Xia	jian 寅	the first month
Shang	jian 丑	the twelfth month
Zhou	jian 子	the eleventh month
Qin	jian 亥	the tenth month

The ancients called this body of rules the san zheng (三正), the three beginnings.

Hu Sanxing, annotating the Zizhi tongjian, states it most clearly:

古有三正：子为天正，周用之，以十一月为岁首；丑为地正，殷用之，以十二月为岁首；寅为人正，夏用之，以十三月（即一月）为岁首。
(There were three beginnings. 子 is the celestial beginning, used by the Zhou, which took the eleventh month as the head of the year; 丑 is the terrestrial beginning, used by the Yin [Shang], which took the twelfth month; 寅 is the human beginning, used by the Xia, which took the thirteenth month — that is, the first — as the head of the year.)

The Shiji, “Treatise on the Calendar”, has a line as well:

夏正以正月，殷正以十二月，周正以十一月。
(The Xia took the first month as their corrected beginning, the Yin the twelfth, the Zhou the eleventh.)

The preceding note, on the double-hour 子, mentioned that the Zhou took the eleventh month as the head of the year and midnight as the beginning of the day. This is the same matter.

Mind the positions of Shang and Zhou: Shang is the twelfth month, Zhou the eleventh. The two are easily reversed, and not a few accounts do reverse them. The text of the Shiji is the standard.

2. The Qin year began in the tenth month

After the First Emperor united the six states, the calendar in use was the Zhuanxu li, and the head of the year was fixed at the month jian 亥 — the tenth month by the Xia reckoning.

So the New Year in Qin was kept in autumn.

One detail must be kept clear: the Qin did not call the tenth month zhengyue. They renamed zhengyue as duanyue (端月), and the pairing of seasons with months continued as before by the Xia reckoning — spring still being the first, second and third months.

A further point. Zhengyue is read zheng yue today. The most widely repeated account is that this avoids the personal name of the First Emperor, 正 and 政 being homophones — the same motive as the change to duanyue. Huang Sheng of the Qing had already rejected that account in his Zigu: the character 正 had a level-tone reading of its own, and was not produced by taboo.

This will matter again shortly.

The early Han went on using the Qin calendar, and did so for more than a hundred years. When Liu Bang took the throne, the New Year ceremony was held in the tenth month. The Shiji, “Biographies of Liu Jing and Shusun Tong”, records that in the seventh year of Emperor Gao of Han the Chang Le Palace was completed and the marquises and ministers all came to court in the tenth month — the palace had just been finished, and the lords and ministers hurried to present their congratulations in the tenth month.

3. Emperor Wu of Han put it back

Under Emperor Wu, in the first year of Taichu, 104 BCE, the matter was settled.

The Taichu li was promulgated that year; it restored the Xia reckoning, jian 寅, and took the first month once more as the head of the year. The work was led by Sima Qian, then Grand Astrologer. Among the more than twenty who compiled the calendar, the calendar officer Deng Ping and Luo Xiahong of Ba commandery contributed most, and the master of methods Tang Du was in charge of the celestial section.

The preceding note, on the twelve double-hours, mentioned the Taichu li as the point at which the twelve took their final form. This is another thing done by the same calendar: it fixed the head of the year as well.

The Taichu li had a longer consequence still. It was the first calendar to incorporate the twenty-four solar terms, and it laid down the rule that a month without a qi is the intercalary month. The skeleton of today’s lunisolar calendar was built then.

From the first year of Taichu, “head of the year” and “zhengyue” became one thing — the first month of the year was zhengyue.

It sounds flat, but it has governed the two thousand years since.

4. After it was changed back, it changed four more times

A change of dynasty required “correcting the first month and changing the colours of dress”, as a sign that the mandate had passed. So after the Taichu li the head of the year moved again several times:

Years	Who	Changed to
8—23 CE	Wang Mang	the Shang reckoning (twelfth month)
237—239	Emperor Ming of Wei, Cao Rui	the Shang reckoning
689—700	Wu Zetian	the Zhou reckoning (eleventh month)
761—762	Emperor Suzong of Tang	the Zhou reckoning

The middle episode is the one with a story.

Wu Zetian was preparing to take the throne. In the eleventh month of the first year of Yongchang (689) she issued an edict adopting the Zhou reckoning, fixing the eleventh month as the first month of the new year, renaming the twelfth month 腊月 and renaming the old first month 一月.

The consequence: the first year of Yongchang had only eleven months, and 325 days.

In that year the people of China lived through a year of 325 days.

Her motive for moving the head of the year was political, not calendrical.

Nine years later (698) she did it again. The calendar officers reckoned that the following year should carry an intercalary twelfth month; what she wanted was for the first day of the first month to fall on “jiazi, new moon at dawn, winter solstice” — the term, the month order and the day-count all beginning on the same day, the hardest of auspices in the eyes of the ancients. She found a pretext, ordered the intercalary month moved elsewhere, forced the day into existence, and at once changed the era name to Shengli.

Modern scholars, back-calculating with the Linde li and astronomical tools, find that the winter solstice of that year fell on the day 壬戌, not 甲子.

The auspice was manufactured.

In the tenth month of the first year of Jiushi (700) she changed it back. The Zizhi tongjian records the reversal bluntly: “On the day jia-yin, an edict restored the first month as the eleventh month, and 一月 as zhengyue.”

5. The head of the year, zhengyue and the seasons are three separate matters

Return to that detail from the Qin.

The Qin took the tenth month as the head of the year, yet did not call the tenth month zhengyue, and ran the four seasons by the Xia reckoning. What does this show?

That for the ancients “head of the year”, “zhengyue” and “which months pair with which seasons” could be three separate matters.

This is not the later reader dividing what the ancients had joined. Sima Zhen of the Tang, annotating the Shiji, writes plainly under the line about the court gathering in the tenth month:

汉以十月为正，故行朝岁之礼，史家追书十月也……案：诸书并云十月为岁首，不言以十月为正月。
(The Han took the tenth month as their corrected beginning, and so performed the ceremony of the new-year court; the historian writes the tenth month after the fact. … Note: the various books all say that the tenth month was the head of the year; none says that the tenth month was zhengyue.)

The tenth month was the head of the year; zhengyue was still zhengyue. More than a thousand years ago, readers of history already kept the two apart.

Keeping them apart, though, is still a matter of how to read old books. One step further down the question changes: does the doctrine of the san zheng itself hold up?

The linguist Wang Li took it as grounds for doubt. If the three dynasties really used three different calendars, why did the Qin not simply rename the tenth month zhengyue?

The reminder is useful when reading old texts.

The Chunqiu, under the first year of Duke Cheng, records that “in the second month there was no ice”. Du Yu’s note on the line runs to six characters: the second month of the Zhou, the twelfth month of today. The second month of the Zhou calendar is the twelfth of the Xia, precisely the coldest time; that the river did not freeze was why the scribe set it down gravely. Read it as the second month of the Xia and the line makes no sense — by then the ice should long since have broken.

Without knowing the difference in the head of the year, one takes the ancients to have made a mistake.

6. A clever proposal that was not adopted

Shen Kuo, in the Northern Song, thought of another way out.

In the Mengxi bitan, “Supplementary Notes”, he proposed setting the moon aside and laying out the months purely by solar terms:

直以立春之日为孟春之一日，惊蛰为仲春之一日，大尽三十一日，小尽三十日，岁岁齐尽，永无闰余。
(Take the day of lichun as the first day of the first spring month, the day of jingzhe as the first day of the second; let the long months run to thirty-one days and the short to thirty; every year then comes out even, and there is never a surplus to intercalate.)

Lichun would be the first day of the first month, jingzhe the first day of the second; the long months would have thirty-one days and the short months thirty, and no intercalary month would ever be needed.

This is a purely solar calendar. It runs almost exactly along the lines of the stem-and-branch calendar — both divide the months by solar terms, and neither needs an intercalary month. As the note on the three calendars said, the stem-and-branch calendar works that way. The difference is that the stem-and-branch calendar never became the everyday calendar of ordinary people, while it has continued in use in charting to this day.

Shen Kuo knew the proposal would draw abuse. He wrote after it:

今此历论，尤当取怪怒攻骂，然异时必有用予之说者。
(This calendrical argument of mine will above all provoke astonishment, anger and attack. Yet another time will certainly find men who use my proposal.)

Abused at the time; sooner or later someone would use it. Read nine hundred years later, the sentence does not sound as though it were addressed to the Song.

Closing

In the course of two thousand years the meaning of the two characters zhengyue has shifted several times. To say “the first day of zhengyue” today points to one thing; to say zhengyue in the Qin points to another month.

What comes next in this series is the clock — the time on a watch is not the double-hour of your birth.

Note on the text

This is a popular exposition, and it makes no claim to original research. The astronomical constants cited are standard values. This English version is issued alongside the Chinese original as the same article in another language. The author’s name is 邝楚嘉, Chujia Kuang; he writes under the pen-name 嘉言一得 (Jiayan Yide).

Declarations

Funding. This research received no external funding.

Competing interests. The author declares no competing interests.

Data availability. No new data were created or analysed in this study.

Language. The original Chinese text is deposited as a companion file to this record (正月本来不是一月).

References

[1] Chunqiu [春秋][M]. Pre-Qin period.

[2] Sima Qian. Shiji [史记][M]. With commentary by Sima Zhen. Western Han.

[3] Sima Guang. Zizhi tongjian [资治通鉴][M]. With commentary by Hu Sanxing. Song dynasty.

[4] Shen Kuo. Mengxi bitan [梦溪笔谈][M]. Song dynasty.

[5] Huang Sheng. Zigu [字诂][M]. Qing dynasty.

Images in this article are illustration only; they are not evidence and not a basis for any statement of fact.

This article is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
You are free to reprint, redistribute, or build upon it, including for commercial purposes, provided you credit the original author as 「嘉言一得（Chujia Kuang 邝楚嘉）」 and include a link to the original publication.
Copyright in the excavated Qin and Han bamboo-slip photographs and in the transmitted texts quoted here remains with their respective rights holders and is not covered by this licence.
For the full terms of the licence, see: https://creativecommons.org
Related articles

• Three Calendars, Three Kinds of Time — the three scales and what each is set to.
• Ten Stems, Twelve Branches: Why Only Sixty Pairings Remain — why ten stems and twelve branches give sixty, not a hundred and twenty.
• Lichun Is Not a Day: The Solar Term Is an Instant — a solar term is an instant, not a day.
• The Gnomon and the Sun at Its Utmost — shadow-measurement before the Dengfeng observatory.
• The Twenty-Eight Mansions on the Lacquer Box from the Tomb of Marquis Yi of Zeng — a lower bound for the four images.

Chinese editions: 正月本来不是一月（the original Chinese text of this essay）

Site hubs: Glossary ｜ Calendar Reform ｜ Publications

All data in this article were collected and compiled by Chujia Kuang.

---

作者：嘉言一得（邝楚嘉）｜独立研究者，致力于可核验的原始文献与历法数据归档。
Author: Chujia Kuang (邝楚嘉), pen name Jiayan Yide (嘉言一得) — independent researcher devoted to archiving verifiable primary sources and calendar data.
平实记录 · 数据集 · 预印本：GitHub / WordPress / Zenodo（DOI: 10.5281/zenodo.22788686 ／ 10.5281/zenodo.23028692）
Records · Datasets · Preprints: GitHub / WordPress / Zenodo
学术身份 · Academic identity：ORCID 0009-0002-7650-833X ｜ OpenAlex A5151908354

最后修订时间：2026年10月18日

天文历法基础 · 第十六篇

文／邝楚嘉

立春在公历上跑不出 2 月 3 日到 5 日这个圈。翻三十年的历书，它就在这三天里换位置。

可要看农历，同一个立春，有时落在腊月，有时落在正月，前后能差出整整一个月。

一套刻度，在两本历书里一个稳如钉子，一个左右摇晃。这件事本身就是二十四节气的来历。

一、一段十五度

国家标准 GB/T 33661-2017《农历的编算和颁行》给节气下的定义很干脆：

二十四节气：一个回归年内 24 个太阳地心视黄经等于 15 度的整数倍的时刻的总称。

关键词是「时刻」。

太阳沿黄道走一圈是 360 度。切成 24 份，每份正好 15 度。太阳走到一个 15 度的刻度上，那一刻就是一个节气。三百六十除以十五得二十四，二十四段就是这么来的。

黄道的 0 度定在春分点。从春分起算，春分是 0 度，往后每个节气加 15 度，到冬至是 270 度，走完一圈回到春分。

有一件事要交代清楚。天文学上从春分起算，民间的节气表却大多从立春起排。两套数法并存，起点不同，名目与顺序完全一样。看表之前先看清它从哪一头起，就不会数错。

二、二十四节气对照表

节气	黄经	公历（大致）	英译
立春	315°	2月3—5日	Start of Spring
雨水	330°	2月18—20日	Rain Water
惊蛰	345°	3月5—7日	Awakening of Insects
春分	0°（360°）	3月20—22日	Spring Equinox
清明	15°	4月4—6日	Pure Brightness
谷雨	30°	4月19—21日	Grain Rain
立夏	45°	5月5—7日	Start of Summer
小满	60°	5月20—22日	Grain Buds
芒种	75°	6月5—7日	Grain in Ear
夏至	90°	6月21—22日	Summer Solstice
小暑	105°	7月6—8日	Minor Heat
大暑	120°	7月22—24日	Major Heat
立秋	135°	8月7—9日	Start of Autumn
处暑	150°	8月22—24日	End of Heat
白露	165°	9月7—9日	White Dew
秋分	180°	9月22—24日	Autumn Equinox
寒露	195°	10月8—9日	Cold Dew
霜降	210°	10月23—24日	Frost’s Descent
立冬	225°	11月7—8日	Start of Winter
小雪	240°	11月22—23日	Minor Snow
大雪	255°	12月6—8日	Major Snow
冬至	270°	12月21—23日	Winter Solstice
小寒	285°	1月5—7日	Minor Cold
大寒	300°	1月20—21日	Major Cold

黄经一栏取正值，所以立春写作 315°，不从负号那一头绕。表里节与气交错排列，单数位是节，双数位是气，这一层往下讲。

英译取的是国际通行的一套，与中国气象局公布的另一套略有出入：气象局把立春写作 the Beginning of Spring，小满写作 Lesser Fullness of Grain。两套都能用，查到哪一套都不算错。

三、它为什么在公历上几乎不动

公历和节气量的是同一个东西，太阳。

公历按回归年编排，节气按太阳在黄道上的位置定，两把尺子对着同一位参照物，刻度自然对得齐。上半年的节多半落在 6 日前后，气落在 21 日前后；下半年的节落在 8 日前后，气落在 23 日前后。年际浮动不过一两天。

农历那边是另一回事。农历月按朔望月定，一个月 29 天半，十二个月比回归年短十一天，攒几年补一个闰月。节气在农历里的日期因此大幅摆动，双春年、无春年都是这么来的。这一层前面讲过，这里只提一句。

四、名字里藏着两批

二十四个名字不是一次写齐的，分成两批看就清楚。

第一批八个：立春、春分、立夏、夏至、立秋、秋分、立冬、冬至。这八个分的是太阳的八个关键节点，两个「立」是四季的开头，两个「分」是昼夜平分，两个「至」是白昼最长与最短。

第二批十六个：雨水、惊蛰、清明、谷雨、小满、芒种、小暑、大暑、处暑、白露、寒露、霜降、小雪、大雪、小寒、大寒。这一批分的是更细的气候变化。

从用字上也能看出两批的来历。暑有一对小暑大暑，雪有一对小雪大雪，寒有一对小寒大寒，露有两处，白露与寒露。雨水、谷雨两个字都带着水。这些字说的都是地上的景象，不是天上的度数。

五、月首叫「节」，月中叫「气」

二十四个名字再分一次，分成两组，各十二个。

节的十二个：立春、惊蛰、清明、立夏、芒种、小暑、立秋、白露、寒露、立冬、大雪、小寒。

气的十二个：雨水、春分、谷雨、小满、夏至、大暑、处暑、秋分、霜降、小雪、冬至、大寒。

一个节配一个气，合成一个农历月。节在月首，气在月中。

落到排盘上，月柱的界线只认节，不认气。立春一到就是寅月，惊蛰一到就是卯月，与农历是不是初一没有关系。这一层前面已经讲过，这里只把两组名目补齐，气的那十二个在排盘里另有用途。

六、气与候：两把尺子，差半拍

历法里有一层容易读岔的地方，得先说明白。

气量的是气机的节律：木气寅月登场、卯月最壮、辰月让位。这把尺子给排盘用，节气一到气就换班，分秒不差。

候量的是物候的体感：正月已建寅，寒气还重，草木不动；卯月才草长莺飞；辰月才真正茂盛。这把尺子说的是地上看得见的样子。

两把尺子差半拍，候比气慢约一个月建。

所以木气到辰月已经衰了，你抬头看地里的树，辰月才刚茂盛，这两句并不打架：一句讲盘上的气，一句讲地上的候。辰、戌、丑、未这四个字特殊，正因为它们落在气已经转弯、候还没有跟上的关口上。

七、「气」这个字有三种用法

同一个字，在本行里至少三个意思，读的时候按语境分。

一是中气，就是跟节对举的那十二个名目，雨水、春分那一组。这是排盘语境里的用法。

二是一个节气的长度，十五天。说「一气」的时候，指的是这一段的日子。

三是分割一年的方法，平气与定气。按时间均分叫平气，按太阳实际位置分叫定气。第三条前面有一篇专门讲过，这里不重复。

八、古人怎么把这二十四段量出来

古人手里没有黄经这个概念。他们靠的是圭表，立一根标杆，量正午的日影。

冬至那天日影最长，夏至最短，两个极点先定下来。有了两极，再往中间细分成二十四段，就有了能用的时间刻度。

把这二十四个名字第一次完整写下来的，是西汉的《淮南子·天文训》。它记的次序从冬至起头：

斗指子，则冬至。

二十四个名字与今天完全一样，这是现存文献里最早的一份完整名单。《史记·太史公自序》里也有一句话带过「二十四节」：阴阳四时、八位、十二度、二十四节各有教令，顺之者昌。

古人量的是日影的长短，今人量的是太阳在黄道上的度数。两套办法对准的是同一组时刻，后一套只是把那把尺子磨得更细。

至于历法出现之前的人怎么过日子，那一段另有专门一篇讲，这里只留一句。

收尾

二十四段排定之后，一年就有了二十四把尺子，每把十五度。

这二十四把尺子最要紧的用处，是把一年切成十二份，一份一个月。月柱换不换，看的是节到没到，与日历翻到哪一页无关。所以排盘之前先查交节时刻，这一步查错，后面全错。

再往后要碰的，是天上的另一套刻度。太阳的位置定年月，月亮的位置定朔望，两套刻度凑在一起，才成了我们今天说的农历。

---

A&Q

问：二十四节气为什么正好是二十四个？

答：太阳沿黄道一年走 360 度，每 15 度落一个节气，360 除以 15 得 24。

问：节气为什么在公历上日期基本固定，在农历里却大幅摆动？

答：公历与节气都以太阳为准，两把尺子对着同一位参照物，所以公历日期年际浮动不过一两天。农历月按朔望月定，一个月 29 天半，与回归年差十一天，要靠闰月补齐，节气落在农历里的日期因此大幅前后移动。

问：节与气有什么区别？

答：二十四个名字分两组，节在月首，气在月中，一个节配一个气合成一个农历月。排盘时月柱的界线只认节，不认气。

问：气与候是一回事吗？

答：不是。气是气机的节律，给排盘用，节气一到就换班；候是物候的体感，说的是地上看得见的样子，比气慢约一个月建。

问：二十四节气的名称最早完整见于哪本书？

答：西汉《淮南子·天文训》。它记的次序从冬至起头，二十四个名字与今天完全一样，是现存文献里最早的一份完整名单。

---

邝楚嘉按：这一篇最该记住的是两组数：15 度和 24。凡遇到节气，先问它落在哪一个 15 度上，再问这一天是节还是气。两个问题问清了，节气这条线就通了。

引用与出处

[1] 淮南子·天文训[M]//淮南子. 刘安 等. 西汉.

[2] 史记·太史公自序[M]//史记. 司马迁. 西汉.

[3] 农历的编算和颁行: GB/T 33661—2017[S]. 中华人民共和国国家质量监督检验检疫总局, 中国国家标准化管理委员会. 北京: 中国标准出版社, 2017.

---

本文图片仅作补充说明，不作证据，也不作事实依据。

本文采用 知识共享署名 4.0 国际许可协议 (CC BY 4.0) 进行授权。
只要在转载、传播或依此进行演绎创作时，注明原作者为「嘉言一得（邝楚嘉）」并附带原始发布链接，即可自由使用（含商业用途）。
文中所引用的秦汉简牍出土照片、传世文献版权归原属方所有，不在本授权范围内。
协议详情参见：https://creativecommons.org

相关文章

• 三本历法·三种时间 —— 阳历、阴历、干支历，三套刻度各以什么为准。
• 立春换岁，还是正月初一换岁？ —— 「换岁」的两套惯例并行了两千多年。
• 立春不是一整天，只有一秒钟 —— 交节是一个定到秒的时刻。
• 正月本来不是一月 —— 「正月」本来指岁首所在之月。
• 节与气：二十四节气的两半 —— 十二个能换月，十二个不能。
• 尧帝派人看星星，是为了种地 —— 《尚书·尧典》里那道观测任命。
• 古人的一天，从晚上十一点开始 —— 子时是一天里离太阳最远的时刻。
• 钟表上的出生时间，不等于你出生的时辰 —— 钟表时间与真太阳时不是同一件事。
• 时间是怎么产生的？ —— 恒星日与太阳日，每天差出近四分钟。
• 天上三条道：黄道、白道、赤道 —— 三条路各管一头。
• 古人的星空：三垣与二十八宿 —— 二十八个星区，天上一套刻度。
• 没有指南针，古人怎么定方向 —— 磁针上船之前，方向怎么定。
• 干支：从一棵树到二十二个字 —— 「干支」原写作「幹枝」，树干与枝条。
• 十个字配十二个字，为什么只剩六十组 —— 阳配阳、阴配阴，一百二十剩六十。
• 木星走一圈不是十二年 —— 岁星纪年的十二只是近似，八十年就露馅。
• 一年为什么被切成二十四段 —— 一段十五度，二十四段一个回归年。

英文文章：Three Calendars, Three Kinds of Time —— the three scales and what each is set to. ｜ Ten Stems, Twelve Branches: Why Only Sixty Pairings Remain —— why ten stems and twelve branches give sixty, not a hundred and twenty. ｜ Lichun Is Not a Day: The Solar Term Is an Instant —— a solar term is an instant, not a day. ｜ The Gnomon and the Sun at Its Utmost —— shadow-measurement before the Dengfeng observatory. ｜ The Twenty-Eight Mansions on the Lacquer Box from the Tomb of Marquis Yi of Zeng —— a lower bound for the four images.

相关枢纽页：历法术语表 ｜ 历法记录：一年从哪一天算起 ｜ 数据集与预印本

本篇全部数据，由邝楚嘉搜集整理。

---

作者：嘉言一得（邝楚嘉）｜独立研究者，致力于可核验的原始文献与历法数据归档。
Author: Chujia Kuang (邝楚嘉), pen name Jiayan Yide (嘉言一得) — independent researcher devoted to archiving verifiable primary sources and calendar data.
平实记录 · 数据集 · 预印本：GitHub / WordPress / Zenodo（DOI: 10.5281/zenodo.22788686 ／ 10.5281/zenodo.23028692）
Records · Datasets · Preprints: GitHub / WordPress / Zenodo
学术身份 · Academic identity：ORCID 0009-0002-7650-833X ｜ OpenAlex A5151908354

最后修订时间：2026年10月18日

天文历法基础 · 第十四篇

文／邝楚嘉

《史记·天官书》里有一句话，写得毫不含糊：岁星「十二岁而周天」。

这句话写在两千年前，此后一千多年，历代治历的人都拿它当准绳。可它不对。

一、这颗星为什么叫「岁星」

五大行星里，木星的脚步不快不慢。水星一年绕太阳四圈多，金星一年多，火星将近两年；木星要十一年多，土星要二十九年多。

木星这个节奏，正好可以当钟用。

古人把周天从西向东等分成十二份，叫十二次。木星差不多一年走一份，走到哪一份，就记「岁在某次」。这颗星因此得名岁星。

十二次的名字，与二十八宿对得上：

次	二十八宿	次	二十八宿
星纪	斗、牛	鹑首	井、鬼
玄枵	女、虚、危	鹑火	柳、星、张
娵訾	室、壁	鹑尾	翼、轸
降娄	奎、娄	寿星	角、亢
大梁	胃、昴、毕	大火	氐、房、心
实沈	觜、参	析木	尾、箕

《史记·天官书》给岁星定的一年行程是「岁行三十度十六分度之七」。

三十度又十六分之七，乘十二，得三百六十五度又四分之一。这正好是古人心目中的周天度数。

所以在古人自己的算法里，「十二岁而周天」是圆得上的，前后不打架。

二、可木星并不照这个走

古人把周天切成三百六十五度又四分之一，是为了迁就太阳——太阳一天走一度，一年走完周天。木星不在这个刻度上。

现代测得的木星绕日一周是 11.862 年，约 4332.6 天。折算到十二次里，木星每过一年走 1.012 次。

比一次多出 0.012 次。

这个零头，单看一年小到测不出来。可它年年攒着。

三、攒够八十年，就露馅了

攒到八十年上下，多出的零头凑满一整次。这一年，岁星本该走到下一个星次，却提前越了过去。

古人给这个现象取了名字：超辰。多迈的那一步，叫「超一辰」。

超一次要多少年，取决于拿哪个周期值去算：

取法	超一次需
按现代的 11.862 年	85.9 年
按通行的 11.86 年	84.7 年
按本库自算的 11.85657 年	82.7 年

三个数都在八十年上下，差别只在取哪一个周期。取哪个都不算出格。古人不认识 11.86 这个数，可他们从典籍的比对里察觉到了不对。

《左传》襄公二十八年留着一句：「岁在星纪而淫于玄枵。」星纪在前，玄枵在后，岁星本该待在星纪，却往前挤进了玄枵。这个「淫」字，说的是越界。

几百年后，刘歆把这件事算实了。《汉书·律历志》里留着他的一句断语：「三十二年，岁在星纪，距辛亥百四十五岁，盈一次矣。」

句中的「辛亥」，是僖公五年正月的朔日。刘歆从那个基准点往下推，推到鲁昭公三十二年，一百四十五年，岁星整整多走了一次。这一句是他逐年推下来的结果。

四、刘歆打了一个补丁

既然发现了，就得修。

刘歆在《三统历》里给岁星换了一条规矩：一百四十四年超一次，也就是一百四十四年走一百四十五次。

这条规矩保存在《汉书·律历志》的三统历术文里：

推岁所在，置上元以来，外所求年，盈岁数，除去之，不盈者以百四十五乘之，以百四十四为法。

用今天的话讲：先拿一百四十五乘，再除以一百四十四，得出的整数叫「积次」，是岁星走过的星次；余下来的叫「次余」，是这一年的零头。从星纪起数，数到哪就是哪。

这个补丁在历法史上是有名的一笔。东汉人边韶在历法讨论里评它两句，收在《续汉书·律历志》：「百四十四岁一超次，与天相应，少有阙谬。」

拿现代的 11.86 年去对，八十五年前后超一次；刘歆说一百四十四年。这个数并不准。

它值得记的地方在别处：这是史书上第一部完整记载下来的历法里，头一回有人把木星的零头写进算法。

五、太岁：造一颗假星来记年

超辰还有个麻烦，出在「看」这件事上。

按十二次推算，岁星该在哪一次，是算出来的；它真在哪一次，得抬头看。算出来的和看到的对不上，纪年的链条就断一节。

于是古人另立了一颗星。这颗星不出现在天上，走法与岁星一样快，方向相反，顺着十二辰一年走一辰。

这颗假星叫太岁。《史记·天官书》里叫它「岁阴」，《淮南子》里叫「太阴」，后来通称太岁。

《史记·天官书》写「以摄提格岁：岁阴左行在寅，岁星右转居丑」，说的就是这一对：真星右转，假星左行，一左一右，互为镜像。

太岁走的十二辰，就是十二地支。太岁的十二个年名，是摄提格、单阏、执徐、大荒落、敦牂、协洽、涒滩、作噩、阉茂、大渊献、困敦、赤奋若。十二个年名配上岁阳十名，就是《尔雅·释天》里那一套。

按太岁定年名，就不必抬头看星了，接着往下数就行。

有一处要分清。这里说的太岁，是那颗假星的走位，属历法；后世说流年时讲的「流年太岁」，借的是同一个词，说的是另一件事。两个「太岁」不能混。

六、最后连假星也不要了

太岁纪年甩掉了观测，可它每十二年在十二辰上转一圈，遇上超辰仍要调整。

真正省事的办法，是换成干支。

十天干配十二地支，六十组一轮，自己转自己的圈，不看天上任何一颗星。不靠观测，就不会因为星走得不准而错位；只要接着数，就能一直往下延。

汉章帝元和二年，公元 85 年，朝廷颁行《元和四分历》。这一次改历，正式废止岁星纪年，改用干支纪年。

从岁星到太岁，再到干支，是一条撤离的路线：先是看着真星记年，接着造一颗假星代替，最后连假星也不要了。

七、十二这个数，有三个来路

一年有十二个朔望月，十二个月加起来三百五十四天多，比回归年少十一天，闰月就是补这十一天。北斗的斗柄指向又把周天分成十二辰，这是地支月建的直接来路。再加上岁星的十二次，凑成三路。

三路里两路看得见天，一路算得出月。让「十二」最终稳下来的，偏偏是那颗星的不可靠。

收尾

古人用木星记年，是因为它走得稳，一眼能看出走到哪儿。等发现它并不那么稳，就换了一颗假星；假星也有麻烦，就换成干支。

六十甲子转起来，跟天上那颗木星再没有关系。十一个零头攒出来的麻烦，到此为止。

再往后讲，要问一个更早的问题：古人眼里的天，到底是什么形状。

---

A&Q

问：木星绕太阳一圈到底多少年？

答：现代测量是 11.862 年，约 4332.6 天。古人按十二年算，一年差出约五十天。

问：什么叫岁星超辰？

答：岁星按十二年一周天推算，每年在十二次里走一次；木星实际走得略快，每年多出约 0.012 次。攒到八十年上下，零头凑满一次，岁星就提前越过本该在的那个星次。这个现象叫超辰。

问：十二次和二十八宿怎么对应？

答：十二次把周天从西向东等分成十二份，每份对应二十八宿里的一到三宿。星纪配斗、牛，玄枵配女、虚、危，一路排到析木配尾、箕。

问：岁星纪年是什么时候废止的？

答：汉章帝元和二年，公元 85 年，颁行《元和四分历》，废止岁星纪年，改用干支纪年。

问：这里的太岁，和民间说的「流年太岁」是一回事吗？

答：不是。历法里的太岁是岁星的镜像，一颗假想出来的星，顺着十二辰走；后世讲的流年太岁借的是同一个词，说的是另一件事。

---

邝楚嘉按：这一段历法史最值得记的，是古人对付误差的三种办法。发现星不准，先修算法；算法补不住，造一颗听话的假星；假星仍麻烦，干脆换成与天象脱钩的干支。三步走下来，「十二」这个数留住了，来路却换了名字。

引用与出处

[1] 左传[M]. 先秦.

[2] 尔雅[M]. 先秦.

[3] 司马迁. 史记[M]. 西汉.

[4] 刘安等. 淮南子[M]. 西汉.

[5] 班固. 汉书[M]. 东汉.

[6] 司马彪. 续汉书[M]. 西晋.

---

本文图片仅作补充说明，不作证据，也不作事实依据。

本文采用 知识共享署名 4.0 国际许可协议 (CC BY 4.0) 进行授权。
只要在转载、传播或依此进行演绎创作时，注明原作者为「嘉言一得（邝楚嘉）」并附带原始发布链接，即可自由使用（含商业用途）。
文中所引用的秦汉简牍出土照片、传世文献版权归原属方所有，不在本授权范围内。
协议详情参见：https://creativecommons.org

相关文章

• 三本历法·三种时间 —— 阳历、阴历、干支历，三套刻度各以什么为准。
• 立春换岁，还是正月初一换岁？ —— 「换岁」的两套惯例并行了两千多年。
• 立春不是一整天，只有一秒钟 —— 交节是一个定到秒的时刻。
• 正月本来不是一月 —— 「正月」本来指岁首所在之月。
• 节与气：二十四节气的两半 —— 十二个能换月，十二个不能。
• 尧帝派人看星星，是为了种地 —— 《尚书·尧典》里那道观测任命。
• 古人的一天，从晚上十一点开始 —— 子时是一天里离太阳最远的时刻。
• 钟表上的出生时间，不等于你出生的时辰 —— 钟表时间与真太阳时不是同一件事。
• 时间是怎么产生的？ —— 恒星日与太阳日，每天差出近四分钟。
• 天上三条道：黄道、白道、赤道 —— 三条路各管一头。
• 古人的星空：三垣与二十八宿 —— 二十八个星区，天上一套刻度。
• 没有指南针，古人怎么定方向 —— 磁针上船之前，方向怎么定。
• 干支：从一棵树到二十二个字 —— 「干支」原写作「幹枝」，树干与枝条。
• 十个字配十二个字，为什么只剩六十组 —— 阳配阳、阴配阴，一百二十剩六十。
• 木星走一圈不是十二年 —— 岁星纪年的十二只是近似，八十年就露馅。
• 一年为什么被切成二十四段 —— 一段十五度，二十四段一个回归年。

英文文章：Three Calendars, Three Kinds of Time —— the three scales and what each is set to. ｜ Ten Stems, Twelve Branches: Why Only Sixty Pairings Remain —— why ten stems and twelve branches give sixty, not a hundred and twenty. ｜ Lichun Is Not a Day: The Solar Term Is an Instant —— a solar term is an instant, not a day. ｜ The Gnomon and the Sun at Its Utmost —— shadow-measurement before the Dengfeng observatory. ｜ The Twenty-Eight Mansions on the Lacquer Box from the Tomb of Marquis Yi of Zeng —— a lower bound for the four images.

相关枢纽页：历法术语表 ｜ 历法记录：一年从哪一天算起 ｜ 数据集与预印本

本篇全部数据，由邝楚嘉搜集整理。

---

作者：嘉言一得（邝楚嘉）｜独立研究者，致力于可核验的原始文献与历法数据归档。
Author: Chujia Kuang (邝楚嘉), pen name Jiayan Yide (嘉言一得) — independent researcher devoted to archiving verifiable primary sources and calendar data.
平实记录 · 数据集 · 预印本：GitHub / WordPress / Zenodo（DOI: 10.5281/zenodo.22788686 ／ 10.5281/zenodo.23028692）
Records · Datasets · Preprints: GitHub / WordPress / Zenodo
学术身份 · Academic identity：ORCID 0009-0002-7650-833X ｜ OpenAlex A5151908354

最后修订时间：2026年10月18日

Chujia Kuang (邝楚嘉)

Independent Researcher, Jieyang, Guangdong, China

Corresponding author: Chujia Kuang, Independent Researcher, Jieyang, Guangdong, China. Correspondence: kuangchujia.com/contact

Abstract

Lichun (立春), the beginning of spring, is ordinarily treated as a calendar day. It is not. It is an instant, defined in the national standard as the moment at which the sun’s apparent geocentric ecliptic longitude reaches an integral multiple of 15 degrees. This note sets out what that definition entails in practice, and shows why the date of lichun nevertheless appears stable to within a day or two in ordinary years. The sixteen years from 2015 to 2030 are then examined: lichun falls on 3 February in four of them, in each case the first year after a leap year. The second half of the note traces the long preference for mean solar terms (平气, pingqi) over true solar terms (定气, dingqi) — proposed by Liu Zhuo in the sixth century, adopted for computation well before it was adopted for printed calendars, and not carried into the calendar itself until 1645. The note is descriptive; it contains no predictive claims.

Keywords: solar terms; lichun; ecliptic longitude; mean and true solar terms; calendar reform; Qing calendar

1. Twelve seconds, and half the notation changes

Take a concrete case.

In 2021, lichun fell at 22:58:39 on 3 February.

One second either side of that moment belongs to two different years and two different months in the stem-and-branch count:

• born at 22:58:38, one second before the term begins: 庚子 己丑 壬午 辛亥
• born at 22:58:50, eleven seconds after it begins: 辛丑 庚寅 壬午 辛亥

The interval between the two is twelve seconds. The day is the same, the double-hour is the same, and the later two designations are identical — but the year designation and the month designation have both changed.

This is not a contrived example. The same situation recurs every year around lichun, lixia, liqiu and lidong, the four terms that begin the seasons.

2. A solar term is an instant

The national standard defines a solar term without ambiguity: it is the instant at which the sun’s apparent geocentric ecliptic longitude equals an integral multiple of 15 degrees.

The operative word is instant.

Ecliptic longitude is the sun’s position measured along the ecliptic. As set out in the preceding note, the sun completes one circuit in a year, and that circuit is divided into twenty-four terms of 15 degrees each. The substance of a solar term is therefore this: the sun has arrived at a particular position.

That moment has no relation whatever to which page the calendar has been turned to. A calendar is made for people to read; the sun does not consult it.

3. Why we remember it as a day

Because the Gregorian calendar and the solar terms both follow the sun. Two scales set against the same body will naturally keep step.

In the first half of the year the terms fall mostly around the 6th and the 21st of the month; in the second half, mostly around the 8th and the 23rd. The spread is a day or two at most.

Keeping step is not the same as being the same thing. A stable date is a consequence, not a cause. The cause is that the sun arrives at that position on time, and that the Gregorian year is itself laid out by the tropical year, so that the two happen to coincide.

Where they fail to coincide, the date moves.

4. Lichun falls on the 3rd in some years, the 4th in others

The moment of each term drifts later every year. The preceding note gave the figure: a few hours accumulate annually, and after some years the accumulation crosses midnight.

Over the sixteen years from 2015 to 2030, lichun falls on 3 February in four of them: 2017, 2021, 2025 and 2029.

The interval is exactly four years, and in each case the year in question is the first after a leap year. February in a leap year carries one extra day, which shifts every subsequent term one step earlier in the calendar.

So the sentence “lichun is on 4 February” is wrong four times in sixteen years.

The 2021 case was especially close: the term fell at 22:58:39, leaving one hour and one minute before the day ended. A little later and it would have dropped to 4 February.

5. The long detour: mean terms and true terms

The earliest scheme divided the year into twenty-four equal intervals of time, each of 15.2 days. This was called the mean term (平气, pingqi). It was easy to compute and easy to lay out in a calendar.

The difficulty is that the sun does not move at a constant rate. The earth’s orbit is an ellipse; it moves fastest in early January and slowest in early July, so the sun’s apparent motion across the sky is correspondingly uneven. The equal-interval scheme assumes a constant rate, and therefore does not fit.

Liu Zhuo, in the sixth century, proposed the more accurate true term (定气, dingqi), fixed by the degree the sun has actually reached.

Yet when the Tang scholars compiled the Dayan li, they added a short note:

凡推日月行度及轨漏交蚀，并依定气。若注历即依恒气也。
(In computing the courses of the sun and moon, and in the calculation of eclipses, the true terms are to be followed. In printing the calendar, the mean terms are still to be followed.)

Eclipses were computed by true terms; printed calendars continued to use mean terms.

To know which was correct and still use the incorrect one is explained by a single consideration: mean terms made the days easy to lay out. The compromise held until the end of the Ming. It was not reversed until the second year of Shunzhi in the Qing, 1645 — an interval of one thousand and forty-one years.

Once true terms were adopted, the intervals between terms ceased to be equal: around the winter solstice the gap is a little over fourteen days, and around the summer solstice it opens to nearly sixteen.

6. So the moment of a term must be recorded to the minute

A term is an instant, and anything determined by a term cannot be recorded to the day alone.

Where the moment falls within an hour or two of the term boundary, the minute is required as well.

This is why old calendars and present-day almanacs both report the moment of a term to the minute, and sometimes to the second.

One further point is easily confused. The moment of a term differs from year to year. When the data for the year in question are needed, the calendars compiled by the Purple Mountain Observatory are the domestic standard; impression is not a substitute. The same term may fall several hours apart in two consecutive years.

7. Closing note

A solar term is an instant, not a day.

As noted earlier, the popular reckoning of the zodiac sign by the first day of the first month and the reckoning of the year by lichun run on two different conventions. This note supplies the second half of that statement: lichun is itself only an instant, and a few minutes either way can mean the difference between two designations of the year.

So when a date is reported without a time, the necessary question is: at what hour and minute?

What follows from this is the double-hour — and the double-hour does not begin at midnight. It is two hours long, and it crosses into the following day.

Note on the text

This is a popular exposition, and it makes no claim to original research. The astronomical constants cited are standard values. This English version is issued alongside the Chinese original as the same article in another language. The author’s name is 邝楚嘉, Chujia Kuang; he writes under the pen-name 嘉言一得 (Jiayan Yide).

Declarations

Funding. This research received no external funding.

Competing interests. The author declares no competing interests.

Data availability. No new data were created or analysed in this study.

Language. The original Chinese text is issued as a separate article; see the Chinese editions below.

References

[1] 大衍历 [Dayan Calendar]. In 旧唐书·历志 [Old Book of Tang: Treatise on the Calendar]. Liu Xu 刘昫 et al. Later Jin. (Lost; reconstructed from quotations.)
[2] 春秋 [Spring and Autumn Annals]. Pre-Qin.
[3] 礼记·月令 [Book of Rites: Proceedings of Government in the Different Months]. In 礼记 [Book of Rites]. Pre-Qin to Western Han.
[4] 淮南子·天文训 [Huainanzi: Treatise on Astronomy]. In 淮南子 [Huainanzi]. Liu An 刘安 et al. Western Han.
[5] 史记·历书 [Shiji: Treatise on the Calendar]. In 史记 [Records of the Grand Historian]. Sima Qian 司马迁. Western Han.

Images in this article are illustration only; they are not evidence and not a basis for any statement of fact.

This article is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
You are free to reprint, redistribute, or build upon it, including for commercial purposes, provided you credit the original author as 「嘉言一得（Chujia Kuang 邝楚嘉）」 and include a link to the original publication.
Copyright in the excavated Qin and Han bamboo-slip photographs and in the transmitted texts quoted here remains with their respective rights holders and is not covered by this licence.
For the full terms of the licence, see: https://creativecommons.org

Related articles

• Three Calendars, Three Kinds of Time — the three scales and what each is set to.
• Ten Stems, Twelve Branches: Why Only Sixty Pairings Remain — why ten stems and twelve branches give sixty, not a hundred and twenty.

Chinese editions: 立春不是一整天，只有一秒钟（the original Chinese text of this essay） ｜ 立春换岁，还是正月初一换岁？ ｜ 正月本来不是一月 ｜ 节与气：二十四节气的两半 ｜ 古人的一天，从晚上十一点开始 ｜ 钟表上的出生时间，不等于你出生的时辰 ｜ 时间是怎么产生的？

Site hubs: Glossary ｜ Calendar Reform ｜ Publications

All data in this article were collected and compiled by Chujia Kuang.

---

作者：嘉言一得（邝楚嘉）｜独立研究者，致力于可核验的原始文献与历法数据归档。
Author: Chujia Kuang (邝楚嘉), pen name Jiayan Yide (嘉言一得) — independent researcher devoted to archiving verifiable primary sources and calendar data.
平实记录 · 数据集 · 预印本：GitHub / WordPress / Zenodo（DOI: 10.5281/zenodo.22788686 ／ 10.5281/zenodo.23028692）
Records · Datasets · Preprints: GitHub / WordPress / Zenodo
学术身份 · Academic identity：ORCID 0009-0002-7650-833X ｜ OpenAlex A5151908354