The Month Pillar Changes at the Exact Solar-Term Minute — Not on the Date

01Why a minute-level boundary exists at all

Solar terms are defined by the sun's apparent position on the ecliptic, so each term is an instant, not a date. The traditional almanac printed them to the minute, and practitioners were expected to read a birth against that printed instant. Modern software inherits the same precision — when it uses it.

The month pillar is the one chart element most visibly affected, because it is also the anchor of the major-luck sequence: its stem and branch determine the starting point of every ten-year period. A birth inside the crossing minute therefore shifts not just one pillar but the entire cycle chain built from it.

02Worked example and near miss: two minutes across 驚蟄

The 2026 example is unusually clean. 驚蟄 falls at 5 March 2026, 21:59:00 Asia/Shanghai. A chart for 21:58 that evening (same day, same hour branch) returns month pillar 庚寅; a chart for 21:59 returns 辛卯. The hour pillar is unchanged (癸亥) because the double-hour boundary is 23:00, one detail that makes the case easy to verify: the only difference is the month, by exactly one position.

The near miss is the day before and after: 4 March and 6 March both fall clearly on one side, and any competent software agrees there. The disagreement is confined to the 21:59 instant and everything inside the minute around it — a window so small that real birth records almost never resolve it, which is precisely why it matters that software states its boundary policy rather than hiding the ambiguity.

A second, more common near miss is the 立春 crossing on the same principle: a birth at 04:00 on 4 February 2026 falls before the 04:02:08 instant and keeps the previous month pillar, while one at 04:03 is in the new month. The pattern is identical; the term is different.

03School variants: instant-based vs date-based month assignment

The dominant convention, and the one used by classical texts and by this engine, assigns the month by the exact instant. Almanac products that print one term per day are compatible with it, provided the printed time is used.

A minority of simple calculators assign the month by the calendar date, effectively treating the term as occurring at 00:00. For most of the year these are indistinguishable; inside a term day they differ for births that happen before the printed instant. This is one of the commonest reasons two apps produce different month pillars for the same date, and it is invisible unless the user was born on a term day — roughly twelve days a year.

A third variant, found in some older almanacs, shifts the boundary to the start of the following day. It is rare in software and is worth checking only when a chart appears off by one month and the birth is on a term day.

04Limits: what the clock and the minute cannot decide

A birth record accurate only to the hour cannot resolve a boundary that falls inside that hour. In that case the honest outcome is two candidate month pillars, and any reading must acknowledge the ambiguity rather than pick one silently.

This page reports engine outputs for stated inputs; it does not claim that one boundary policy produces a factually better life reading. The month pillar's job in the chart is to carry seasonal strength and to anchor the major-luck sequence, and a one-position error propagates through both.

The month pillar also does not fix the lunar month. A leap month (閏月) affects only the lunar calendar's numbering, never the stem-branch month — see the solar-terms-and-calendar page for that distinction.

05Checklist

• The birth date was checked for whether it falls on a sectional-term day
• The exact term instant (to the second) was retrieved, not just the date
• The birth time was compared against that instant on the same day
• If the birth record is hour-precise only, both candidate month pillars were acknowledged
• The software's stated boundary policy (instant vs date) was recorded with the chart