# China's Clock Ran an Hour Fast for 3,262 Days A day-by-day scan of 1901 to 2099 finds 14 windows when China's clock ran an hour ahead: 3,262 days, only 27% of them in 1986 to 1991. - Canonical: https://gptbazi.com/blog/clock-an-hour-fast-china - Language: English - Published: 2026-11-04T17:40:00+08:00 - Updated: 2026-11-04T17:40:00+08:00 - Author: Ziheng — https://gptbazi.com/author/ziheng > GPTbazi is a personality map for entertainment and self-reflection. It is not medical, legal, financial, or psychological advice, and it is not a prediction of events. This article is offered for reflection and entertainment only. Any pattern described here is a reading of a chart, not a verdict about the person named, and nothing on this page is advice or a forecast of events. ## The six years everyone checks are 27% of the record Almost every guide to Chinese birth-time errors names the same six years: 1986 to 1991, when the country ran daylight saving time. This site is no exception. The article on Mbappé's two birth times gives those dates correctly, and so does the piece on solar time. The scan says those six years are 27% of the problem. Fourteen separate windows in the Asia/Shanghai record sit at an offset other than +08:00, totalling 3,262 days out of the 72,684 scanned between 1901 and 2099. The 1986 to 1991 block is 896 of those days. The rest is older and much larger. The two pre-1949 blocks add up to 2,366 days, which is 2.64 times the block everybody checks. So the first correction here is arithmetic rather than insight: if you assume the only clock problem is a summer in the 1980s, you are looking at less than a third of the record. ## What the scan measured, and what it did not Every day from 1 January 1901 to 31 December 2099 was read at 12:00 UTC against the time-zone database that ships with Node, one offset per day. Days whose offset was anything other than +08:00 were merged into runs, and the fourteen windows are simply those runs. The script is in this site's repository and a full re-run takes seconds. Two limits belong up front. The database records legal time, not the clock on the wall of a village with no radio. And a run is counted in calendar days, so a window that opens on 4 May includes 4 May itself. What the scan does establish is narrower and firmer than memory: on these dates the official offset was nine hours ahead of UTC, so a birth written down as 08:00 was written while the sun and the standard meridian were an hour apart. That is a statement about a record, and it can be checked by anyone with the same script. ## Fourteen windows, three eras All fourteen sit in three eras. 1919 holds one, from 13 April to 30 September, 171 days. The 1940s hold seven, running from 1 June 1940 to 27 May 1949 and adding up to 2,195 days. The 1986 to 1991 block holds six, each opening in mid-April or early May and closing in mid-September, adding up to 896 days. The longest single window is not seasonal at all. From 31 January 1942 to 1 September 1945 the offset stayed at +09:00 for 1,310 consecutive days, straight through four winters. Anyone who assumes these were summer adjustments will mis-date the largest block in the table. The three era counts are 171, 2,195 and 896, and they add to 3,262, the total. Those are the numbers to check before trusting any sentence below them. ## Which window you are in Four cases, and only one of them really needs the table. Born before 1949? Read the 1919 row and the seven rows of the 1940s, then ask a second question the table cannot answer: was the birth in a coastal city that kept official time, or somewhere the local clock followed the sun. Born between 1986 and 1991 in April, May or September? The window is narrow and the date is easy to check, because these boundaries are exact days rather than months. Born from 1992 onwards? The record has no irregular offset at all in those years, so the hour question is settled in advance: the clock and +08:00 were the same thing. If none of these fits, the answer is usually the year on your ID rather than the story your family tells. ## Why the clock ran an hour ahead Three unrelated decisions produced the same offset. In 1919 it was the foreign concessions in Shanghai, which adopted summer time while most of the country did not. From 1940 to 1949 it was war and its aftermath, which is why the 1942 window runs through the winter instead of closing in September. From 1986 to 1991 it was a national energy policy, announced each spring and ended each autumn. The mechanism is identical in all three: the legal clock moves forward sixty minutes while the sun, the meridian and every traditional calendar calculation stay where they were. Nothing about the birth changes. Only the label on the time changes. That is why the error stays silent. Nobody remembers being told to move a clock. They remember a time, usually repeated by a relative, and the time they remember is the one printed on the certificate. ## An hour is a whole pillar, half the time The Chinese hour is a two-hour block, twelve to a day, with boundaries on the odd hours: 23:00, 01:00, 03:00 and so on. Shift a birth time by sixty minutes and the hour pillar changes only if the original time sat inside the hour before a boundary. With birth times spread evenly across a day, that is roughly half of them. So the honest version is not that an hour changes your chart. It is that half the people inside these windows have an hour pillar that is wrong, and the other half have one that is right for the wrong reason. There is a second line in the same neighbourhood, and this page does not cross it: whether the day pillar turns at 23:00 or at midnight is a separate dispute between schools. The point to carry away is that one birth time can carry two questions at once, and daylight saving is only the one asked first. ## Check your own date in four steps Step one is to write down the date and clock time as recorded, not as remembered. Step two is to look that date up against the fourteen windows; anything outside them was +08:00 and needs no adjustment at all. Step three applies only inside a window: subtract sixty minutes from the clock time and write that down as standard time. Step four is to cast the chart twice, once from each time, and see whether the hour pillar moved. Step four is the one people skip, and it is the only one that pays. If both charts give the same hour pillar, daylight saving was academic for you. If they differ, you now know which chart your reading has been using all along. For births outside mainland China this table does not apply, because every zone has its own history. The same script can be pointed at another zone and will produce the same kind of list. ## Where the table is silent or wrong Four gaps are worth naming. The table counts calendar days, so the first and last day of a window are included; a birth on a boundary day should be treated as one hour of uncertainty rather than a fact. It covers 1901 to 2099 and says nothing earlier, which rules out most grandparents. It says nothing about the precision of the record, and a certificate reading 03:00 often means around three. The fourth gap is local practice. A birth in a county that kept local sun time in 1943 is not described by a database entry for Shanghai, however exact that entry is. The table answers one question only: whether the legal clock was fast. It does not answer whether the clock in the room was accurate. The two can add up, and they can cancel out. ## What this scan cannot tell you The 1940s rows are the weakest part of the table, and they are also the largest part. The database keeps a single Asia/Shanghai history, and for those years that history is Shanghai's. Whether the northwest and the border regions actually ran on +07 or +08 in a given year is a question the database does not answer, and this page will not pretend to. Settling it needs local sources: a county gazetteer, the advertisement column of a newspaper, a railway timetable. Two sources that disagree are more useful than one that sounds confident. So the honest form of this article's claim is narrower than the headline: 3,262 days of official clock irregularity exist in the record, and the 1940s share is the part I trust least. What I trust most is the 1986 to 1991 block, where the dates are recent, short and easy to verify. ## Where this reading stops This is a time-zone audit, not a forecast. It does not say what a chart means, whether a reading is accurate, or what will happen to anyone, and the site's disclaimer applies in full. Two things can move the numbers above. The time-zone database is revised a few times a year, and a corrected 1940s entry would shift a day or two. And any future legislation bringing summer time back would add windows after 2099, which the next run of the script would pick up. A page that makes a numerical claim should be checkable by a reader who does not trust it. The way to make that possible is to publish the scan next to the conclusion, which is why the script, the raw file and the day count all sit in the open here. That is the practical reason to keep the script rather than the paragraph: the answer here is not a sentence someone wrote once, it is a number that can be recomputed, compared and corrected. ## Key points - A day-by-day scan of the Asia/Shanghai record from 1901 to 2099 finds 14 windows at +09:00, totalling 3,262 of 72,684 days. - The 1986 to 1991 block is 896 days, or 27% of them; the two pre-1949 blocks add up to 2,366 days, 2.64 times as much. - The longest single window ran 1,310 consecutive days, from 31 January 1942 to 1 September 1945, through four winters. - Chinese hours are two hours wide with boundaries on odd hours, so shifting a time by sixty minutes moves the hour pillar for about half of all births. - The table records legal time only; it cannot say whether a 1943 birth in a border county ran on +07 or +08. ## Sources - IANA Time Zone Database: the record this scan reads, shipped with Node as ICU data — https://www.iana.org/time-zones - Node.js Intl.DateTimeFormat, the API used to read each day's offset — https://nodejs.org/api/intl.html - timeanddate.com: time zone and clock changes for Beijing, including 1986 to 1991 — https://www.timeanddate.com/time/zone/china/beijing - Wikipedia: Time in China (Chinese Wikipedia, 中国时区, historical daylight saving sections) — https://zh.wikipedia.org/wiki/%E4%B8%AD%E5%9B%BD%E6%97%B6%E5%8C%BA - Wikipedia en español: Hora de China — https://es.wikipedia.org/wiki/Hora_de_China - Wikipedia 日本語版: 中国標準時 — https://ja.wikipedia.org/wiki/%E4%B8%AD%E5%9B%BD%E6%A8%99%E6%BA%96%E6%99%82 - GPTbazi BaZi methodology (hour pillar, day boundary, solar time) — https://gptbazi.com/about#method --- Plain-text version of https://gptbazi.com/blog/clock-an-hour-fast-china — same content, no scripts. Rendered page: https://gptbazi.com/blog/clock-an-hour-fast-china