How Do You Know a BaZi Calculator Got Your Birth Time Right?
三个出生时间 · 一份可复现的验收集
Test it. Converting clock time to true solar time takes exactly three terms — historical daylight saving, distance from the time-zone meridian, and the equation of time — and each has a checkable right answer. Below are three birth times, the full arithmetic, and the hour pillar a correct engine must return. It takes about ten minutes.
A word on “right,” because it usually means “agrees with my school.” It doesn’t here. What a chart means is a cultural tradition you can take or leave. The input layer is arithmetic: given a date, a clock time, a time zone and a longitude, there is one correct local apparent solar time, checkable to the minute by anyone. Everything on this page is computed at build time by the same modules that power the Auspice calculator; the background on the correction itself is on the true solar time page.
What has to be added up
Historical daylight saving. Not today’s rules — the rule in force on the birth date. Mainland China ran a now-forgotten DST experiment from 1986 to 1991; a Beijing birth certificate from that window reads an hour ahead of standard time, and no present-day time-zone dropdown will tell you so.
Longitude. Solar time moves 4 minutes per degree away from the zone’s standard meridian. China spans five geographic zones on one clock; Vancouver sits at 123°W in a zone whose meridian is 120°W. This term is constant for a given place, which is why it’s the one most often implemented.
The equation of time. The sun itself runs up to about −15 to +16 minutes fast or slow across the year, from orbital eccentricity and axial tilt. It gets dropped most often, on the assumption that it’s a rounding error. It is up to a quarter of a two-hour branch, and in case 3 below it is the entire difference between two charts.
The three test cases
Each case gives an exact input. Feed it into whatever calculator you use and compare the hour pillar that comes back. (Gender is set to female throughout; it affects only the direction of the luck cycles, never the four pillars.)
Case 1 — Beijing, 15 June 1988, 07:30
Tests: Historical daylight saving — China observed DST from 1986 to 1991
Input: 1988-06-15 07:30 · Asia/Shanghai · 116.41°E
| Component | Minutes |
|---|---|
| Daylight saving | −60.00 |
| Longitude (116.41°E vs the zone meridian) | −14.36 |
| Equation of time | −0.45 |
| Total correction | −74.81 |
Local apparent time 06:15 — Mǎo 卯 (Rabbit hour), hour pillar 辛卯 Xīn Mǎo.
What the partial implementations return: 07:30 with no correction at all (壬辰 Rén Chén), 07:15 with longitude only (壬辰 Rén Chén), 06:15 with longitude and daylight saving (辛卯 Xīn Mǎo). The last of those already reaches the right branch — this case is decided by the daylight-saving term.
Case 2 — Vancouver, 20 July 1995, 12:00
Tests: Daylight saving plus the west edge of a time zone — and what it does downstream
Input: 1995-07-20 12:00 · America/Vancouver · 123.12°W
| Component | Minutes |
|---|---|
| Daylight saving | −60.00 |
| Longitude (123.12°W vs the zone meridian) | −12.48 |
| Equation of time | −6.01 |
| Total correction | −78.49 |
Local apparent time 10:41 — Sì 巳 (Snake hour), hour pillar 乙巳 Yǐ Sì.
What the partial implementations return: 12:00 with no correction at all (丙午 Bǐng Wǔ), 11:47 with longitude only (丙午 Bǐng Wǔ), 10:47 with longitude and daylight saving (乙巳 Yǐ Sì). The last of those already reaches the right branch — this case is decided by the daylight-saving term.
Case 3 — Ürümqi (Beijing clock), 3 November 2001, 09:00
Tests: The equation of time, which here moves the answer back across a boundary — and a birth time too close to that boundary to call
Input: 2001-11-03 09:00 · Asia/Shanghai · 87.62°E
| Component | Minutes |
|---|---|
| Daylight saving | 0.00 |
| Longitude (87.62°E vs the zone meridian) | −129.52 |
| Equation of time | +16.35 |
| Total correction | −113.17 |
Local apparent time 07:06 — Chén 辰 (Dragon hour), hour pillar 庚辰 Gēng Chén. And only 6 minutes from the next boundary — see below.
What the partial implementations return: 09:00 with no correction at all (辛巳 Xīn Sì), 06:50 with longitude only (己卯 Jǐ Mǎo), 06:50 with longitude and daylight saving (己卯 Jǐ Mǎo). The last of those is still wrong — this case is decided by the equation of time.
The divergence table
Same three birth certificates, four levels of implementation, hour pillar returned:
| Implementation | Case 1 | Case 2 | Case 3 |
|---|---|---|---|
| Clock time as recorded | 壬辰 Rén Chén | 丙午 Bǐng Wǔ | 辛巳 Xīn Sì |
| Longitude only | 壬辰 Rén Chén | 丙午 Bǐng Wǔ | 己卯 Jǐ Mǎo |
| Longitude + daylight saving | 辛卯 Xīn Mǎo | 乙巳 Yǐ Sì | 己卯 Jǐ Mǎo |
| All three terms | 辛卯 Xīn Mǎo | 乙巳 Yǐ Sì | 庚辰 Gēng Chén · uncertain |
Every row above the last is a chart somebody is currently walking around with. None of these are exotic inputs: a mainland birth during the DST years, a diaspora birth on the west edge of a time zone, and a birth in a province that has kept Beijing time since 1949.
What a wrong hour actually costs
It doesn’t stop at one pillar. Case 2 run all the way through, uncorrected against corrected:
| Uncorrected (12:00) | Corrected (10:41) | |
|---|---|---|
| Hour pillar | 丙午 Bǐng Wǔ | 乙巳 Yǐ Sì |
| Day master | 壬 Yang Water | 壬 Yang Water |
| Strength verdict | Weak | Strong |
| Favourable elements | Metal, Water | Earth, Fire |
| Symbolic stars (神煞) | 3 | 5 — adds Nobleman (Hour), Travelling Horse (Hour) |
| Luck cycles start | 2001-10-30 | 2001-11-09 |
The strength verdict inverts, and the favourable elements invert with it — the uncorrected chart recommends the elements the corrected chart says to avoid. Nearly every downstream conclusion, in any school, is built on that one line. That line rests on 78 minutes of arithmetic.
When the honest answer is "we can’t tell"
Case 3’s corrected time is 6 minutes past a boundary. That is inside the noise floor of a human-recorded birth time — very few records are trustworthy to 6 minutes, and the equation-of-time approximation itself is only good to about one. So the honest output is not 庚辰 Gēng Chén. It is both charts, labelled: shift the recorded clock by 7 minutes and you are on the other side.
| Chén 辰 hour | Mǎo 卯 hour | |
|---|---|---|
| Hour pillar | 庚辰 Gēng Chén | 己卯 Jǐ Mǎo |
| Year, month, day pillar | identical | |
| Day master, strength, favourable elements | identical | |
| Luck cycle stems and branches | identical | |
| Extra marker | — | Peach Blossom (Hour) |
| Luck cycles start | — | 10 days later |
That table is more useful than a confident wrong answer, and it costs one extra call to a pure function. Auspice raises this flag whenever the corrected time lands within 8 minutes of a boundary.
How to test the calculator you use
- Enter case 1. If the hour pillar comes back 壬辰 Rén Chén, historical daylight saving is not being applied — the most common failure, and the easiest to check.
- Enter case 2. 丙午 Bǐng Wǔ is the same diagnosis; 乙巳 Yǐ Sì means daylight saving and longitude are both in. Then look at whether the tool gives you a strength verdict, and compare it against the uncorrected run.
- Enter case 3. 己卯 Jǐ Mǎo means the equation of time is missing; 庚辰 Gēng Chén means all three terms are in.
- Check whether case 3 is flagged as near a boundary. A tool that hands you one pillar with no caveat is asserting a precision the input data does not have.
- Look for the breakdown. A tool that shows the three components as an addition table can be audited. A tool that shows one corrected time, or nothing at all, is asking you to take its word. A useful negative check: if the time-zone selector holds a handful of entries instead of IANA zones, historical rules cannot be right — "GMT+8" is not enough information to know whether it was 1988 in Beijing.
Where correct implementations may still differ
Not every disagreement is a bug, and a test that can’t tell the difference isn’t worth much. The late Zi hour (23:00–24:00) is assigned to the current day by one school and the next day by another — a real convention split, not an error; what is an error is not telling you which one you got. Equation-of-time approximations differ by well under a minute, and a city centroid versus a hospital address is a minute or two; both matter only right at a boundary, which is why the boundary flag exists. Our conventions are listed on the method page. None of this explains an hour of difference — an hour of difference is a missing term.
Why publish a test we could fail
Because the alternative is asking people to trust a black box, in a field that has earned every bit of its credibility problem. What a chart means is a tradition; whether the chart was computed correctly is not a matter of belief. Calendar conversion, historical time-zone resolution, solar position and the sexagenary cycle have right and wrong answers. These three cases run against our engine on every build — this page is generated from their output, so if the engine ever changed, the numbers here would change with it. If you find an input where we are wrong, that is a bug report and we want it: contact, and corrections get logged publicly on the corrections page.
Run your own birth data through the calculator — free, no account. It shows the correction broken into the same three components, flags the 8-minute boundary case and casts the neighbouring chart, and if you don’t know your birth hour it computes all twelve candidates and leaves blank whatever they disagree on. Background on the correction itself: true solar time in BaZi. Auspice is for reflection, not prediction.
Every number on this page is computed at build time by lib/bazi — the same modules behind the calculator — from the inputs shown. Nothing here is transcribed by hand.