China's 1986–1991 Daylight Saving Time: The Hour Pillar Shift Nobody Checks

01The six years, and why they are easy to miss

China reintroduced daylight saving time in 1986 and suspended it after 1991. Each year the clocks advanced in mid-April and returned in mid-September; the exact dates varied, and in 1990 the advance was on 15 April and the return on 16 September. Births inside those windows appear on certificates in daylight time, and nothing on the document says so.

The effect is invisible for most of the year and for most of the day: a one-hour shift changes the hour pillar only when the birth falls within an hour of a two-hour boundary, and changes the day pillar only near midnight. That combination — six years out of a century, a few hours of the day — is why the issue survives in production calculators.

02Worked example and near miss: 03:30 on the changeover morning

On 15 April 1990 the clocks jumped from 02:00 to 03:00. A birth recorded at 03:30 is therefore 02:30 in standard time. The true-solar-time correction for Beijing on that date is about −58 minutes (equation of time plus the DST removal), placing the birth around 01:32 — the Chou hour (丑), pillar 丁丑. A calculator that ignores the historical offset reads 03:30 as standard time, subtracts only the equation of time, and lands near 02:32 — still inside the Chou hour here, but within minutes of the Yin boundary.

The cleaner near miss is 23:30 on a summer date. Under daylight time that is 22:30 standard; a chart that skips the correction treats it as 23:30 and may apply the late-Zi rollover, moving the day pillar by one. The hour pillar also differs: Hai (亥) versus Zi (子). Two conventions, one hour apart, two different charts.

03School variants: who corrects and who does not

Software that reconstructs local time from the IANA historical database — as this site's engine does — applies the offset automatically for the birth date. Most consumer calculators apply a fixed UTC+8 and never consult history; for dates outside 1986–1991 the two agree, which is why the bug is rarely reported.

A third group applies DST only when the user manually flags it. That is better than nothing but depends on the user knowing the birth fell in a DST window.

Outside mainland China the same issue appears with different dates: Korea used daylight time in 1948–1960 and 1987–1988, the United States changed its DST rules several times, and some regions skipped or doubled an offset entirely. The principle is identical; only the table differs.

04Limits: what history cannot settle

A certificate records civil time, but not whether the family already adjusted it, and not whether the birth was recorded before or after the changeover hour. Where the record is ambiguous, the honest output is two candidate charts.

Historical time-zone data itself has versions. Citing a transition requires stating the database and its retrieval date, exactly as this page does.

DST correction does not replace the true-solar-time correction, and it does not resolve the 23:00 rollover question. All three interact on the same night, and each must be applied in order: historical offset → longitude and equation of time → day-rollover convention.

05Checklist

• The birth year was checked against China's 1986–1991 DST windows
• The birth date was checked against that year's exact start and end dates
• One hour was subtracted before any hour-pillar or rollover calculation
• The birthplace's time zone was checked for its own historical DST rules if outside mainland China
• The time-zone database version was recorded with the chart