True Solar Time in BaZi

真太阳时 · 钟表在撒谎，太阳没有

Yes. In Chinese astrology the twelve two-hour blocks of a BaZi day were defined by the sun, not the clock. Casting a chart means converting clock time to true solar time (真太阳时) — daylight saving, longitude, equation of time — which can change your hour pillar.

Why does the clock differ from the sun?

Time zones are a railway-era convention: whole regions agree to pretend they share one noon. The earth turns 15° of longitude per hour, so every degree you stand west of your zone’s reference meridian puts the sun 4 minutes behind the clock. On top of that, the earth’s tilted axis and elliptical orbit make the sun run fast or slow through the year — the equation of time, which swings between about −15 and +16 minutes. And if daylight saving was in force, the clock was moved a further hour off the sun on purpose.

True solar time = clock time − daylight saving − distance from the meridian + equation of time. The three parts always add up exactly to the total correction, so the arithmetic can be checked line by line.

How big is the correction in your city?

Computed with the same solar-time module that powers the Auspice calculator (IANA historical time zones, NOAA equation-of-time approximation), for noon on 1 January and 1 July 2000. The two columns differ where daylight saving applies — note Sydney and São Paulo flip seasons.

Minutes to add to clock time. The remainder between the longitude part and the total is the equation of time plus, where active, daylight saving (−60).

Two worked examples. A 2:30 pm birth in Vancouver on 15 July 1990 is 13:11 by the sun (-78 min: -60 daylight saving, -12 longitude, -6 equation of time) — enough to leave the Goat hour for the Horse hour. And an 8:00 am birth in Ürümqi the same day is 4:44 by the sun (-195 min), because the recorded clock was Beijing time — and 1990 China was under its brief daylight-saving experiment.

When does the correction actually change a chart?

More often than the folk wisdom says — it depends almost entirely on where. The two-hour blocks flip at odd clock hours (1, 3, 5…), and a correction only matters if it walks the time across one of those boundaries; the next section counts how often that happens, city by city. Our calculator flags an hour pillar as uncertain when the corrected time lands within 8 minutes of a boundary — in that case, read both charts rather than trust either. The people most likely to cross a boundary: births under daylight saving, births far from the time-zone meridian (western China, most of Spain, parts of Canada and Brazil), and recorded times that were rounded to the hour. If you want to check whether the calculator you use applies all three terms, there are three test birth times with a checkable right answer for each.

How often does the correction change the hour pillar? A 40-city count

We took 40 cities, four reference years (1980, 1990, 2000, 2010) and every clock minute in each — 2,102,400 minutes per city — converted each to true solar time with the same arithmetic as above, and counted how many landed in a different two-hour block than the clock said. That fraction is the chance that a birth at a random clock minute in that city carries the wrong hour pillar if nobody corrects it. Minutes are weighted equally (births cluster in daytime, so this is a per-minute figure, not a per-baby one); month and year pillars, which flip only at solar-term instants, are not counted.

Result: in the main overseas-Chinese birth cities the hour pillar changes for about 40% of clock minutes; across mainland China, all on Beijing time, 48%; elsewhere in Asia 22%. Toronto 44%, Vancouver 40%, London 33%, New York 29%; Madrid 84%; Kashgar 100%. Taipei, closest to its meridian, 7%.

Top 20 of 40 cities by hour-pillar change rate, averaged over 1980/1990/2000/2010. Mainland Chinese cities show a DST column only because 1990 fell inside China's 1986–1991 daylight-saving years. Full table for all 40 cities and the script that produced it: data/solar-time-study.json in the Auspice repository.

Three things the table says that the usual advice does not. First, daylight saving is the main culprit in the West: Toronto’s rate is 64% under DST and 17%without it, so a summer birth in Canada is closer to a coin flip than to a safe bet. Second, “only overseas births need correcting” is wrong: China runs one clock across five time zones of longitude, so Chengdu sits at 58% and Ürümqi at 99% while Shanghai, near the 120°E meridian, is 11%. Third, Spain is the European outlier (84%) because it keeps Berlin’s clock on Lisbon’s longitude. Singapore (48%) is the quiet surprise: no DST, but a UTC+8 clock on a UTC+7 longitude.

Which daylight-saving rules does the chart use?

Historical ones, not today’s. Daylight-saving rules change constantly, and a chart must use the rule in force on the birth date: mainland China observed DST from 1986 to 1991, Taiwan and Hong Kong had early periods of it, São Paulo kept it until 2019. We resolve birth times against the IANA time-zone database (via ICU), which encodes this history — the Ürümqi example above only comes out right because 1990’s rule differs from 2026’s.

The hour that never existed

When a clock springs forward, the skipped hour never happened locally. On April 15, 1990 in mainland China the clock jumped straight from 01:59 to 03:00, so a birth recorded as 02:30 that night names a time that did not exist — and the same is true on 6 spring-forward dates between 1986 and 1991. Most calculators resolve such an input silently, which is how two adjacent entries (02:59 and 03:00) can end up with different hour pillars despite naming the same instant. We detect it instead: the chart resolves to the moment the clock jumped, and the reading is marked as resting on a time that never occurred. The likely explanation is that the family read a clock not yet moved forward — meaning roughly 01:30 — but a record cannot settle which, so the honest answer is two charts, not one.

What about births between 11 pm and midnight?

The Zi hour (子时) runs 23:00–01:00, straddling midnight, so the 23:00–24:00 slice must be assigned to a day by convention. We use the same-day convention (sect 2): your day pillar stays with the date you were born on. Another long-standing school rolls it to the next day, and no arithmetic settles the dispute — a birth in that hour genuinely has two defensible charts. The full convention list is on the method page.

What if the birth time is unknown?

Three of the four pillars come from the date alone, so an unknown hour costs you the hour pillar and nothing else — the day pillar, the one that means “you,” is safe. The Auspice calculatorcasts unknown-hour charts from what the twelve candidate hours share, and treats a remembered “around dinnertime” as the window it really is.

Every number on this page is reproducible: the exact weights and conventions are documented on the method page, and your own chart shows its correction broken into the same three parts — free, no account. For the wider system, read what BaZi is. Auspice is for reflection, not prediction.