The history of BaZi: from oracle-bone day counts to the four pillars

In short

BaZi, the four pillars, is a separate story from the Babylonian, Greek and Indian horoscope. It takes no planetary positions at all: its inputs are calendar labels from the cycle of 10 stems and 12 branches. Its layers are of very different ages: the day count is over 3,000 years old, the year label is Han, the month and hour stems are worked out by rule, and the Day Master method is Song or later.

The 60-day count

The 60-term cycle of stems and branches appears on the late-Shang oracle bones, from the reign of Wu Ding (about 1250 to 1200 BCE), where it is used only to count days. Eclipses recorded in the Spring and Autumn Annals (722 to 481 BCE) show that the day count has run unbroken since then. Stem-branch years first appear in the Mawangdui silk manuscripts, from a tomb sealed in 168 BCE, and became general after the calendar reform of 85 CE. Months and the twelve double hours were named by branches alone; their stems are derived from the year and day stems by rhyming rules that Ming writers already called old verses. Of BaZi's four pillars, only the day pillar is an ancient continuous count.

Sky-reading for the state

Early Chinese astronomy, like Babylonian, read the sky for the state. Field allocation (fenye) mapped stars to regions so an omen could be placed on the ground, and Sima Qian wrote the Treatise on the Celestial Offices around 100 BCE as court astronomer. The 28 lodges (xiu), of unequal widths, are first listed in full on a lacquer chest from a tomb sealed after 433 BCE. Whether they share an origin with India's nakshatras is still debated; when Buddhist translators compared the two in the 8th century, most did not correspond.

From three pillars to four

Fate calculation from birth dates probably began in the late Han, according to the historian Chao Wei-pang (1946), but all manuals before the Tang are lost. The first well-documented practitioner is the Tang official Li Xuzhong (762 to 813). His epitaph, by Han Yu, says he worked from the stems and branches of the year, month and day of birth, with no hour.

The method used today, which takes the stem of the birth day as the person (the Day Master) and adds the hour to make four pillars and eight characters, is ascribed to Xu Ziping, a 10th-century figure about whom little is known, and was written down by Xu Dasheng in the late Southern Song. Chao treats them as two people centuries apart. Its relations, such as which element feeds or controls the Day Master, rest entirely on the production and control cycles of the five phases. The standard texts are later still: the Yuanhai Ziping (probably a Ming re-edition of Xu Dasheng's work) and Wan Minying's San Ming Tong Hui (16th century).

Five phases and the solar terms

The five phases (wuxing) appear in Zhou texts and were systematised with yin and yang by Zou Yan in the 3rd century BCE, becoming the organising scheme of Han thought. The 24 solar terms entered the official calendar with the Taichu reform of 104 BCE. In BaZi the year changes at Lichun and the month at each of the twelve jie terms; that is the one place the Sun's real position enters. When fate calculators first adopted those boundaries is not documented in the sources we found.

When the horoscope reached China

Indian astrology reached China with Buddhist translations, above all Amoghavajra's Xiuyao jing (759, revised 764), which also brought the seven-day week. The horoscope itself, with planets and an ascendant, arrived around 800 with Iranian translators: a version of Dorotheus, the Greek astrologer of the 1st century, brought to China between 785 and 805. That planet-based tradition stayed separate from BaZi; Ming writers distinguished the school of Ziping from the school of the five stars. No source shows that the foreign horoscope caused the Tang turn to birth-date fate calculation.

How Jyotishya uses this

Jyotishya computes the four pillars by the calendar rules (the year from Lichun, months from the jie terms, local mean solar time for the day and hour) and scores each day by the Day Master method in a simplified fixed form. Knowing that the Day Master is a Song-era method, and that the day count is far older, is part of showing what the rule is. Nothing here suggests the rules predict anything.