Birth Time, Time Zones, DST, and Accuracy

The astronomy in a birth chart is not the fragile part — that pipeline is written out step by step elsewhere, and its worst approximation is worth a quarter of an arcminute on the worked example there — one chart, one latitude, a recent birth. That page says plainly that its figures do not transfer unchanged, and an older birth carries a larger obliquity term. The fragile part is the number you typed into the time box. An hour of error moves the Sun 0.04°, the Moon about half a degree, and the rising sign about fifteen — some 366 times the Sun's figure. Swept across 33,966 sample birth moments, a birth time that is off by one hour lands on a different rising sign 49% of the time. This page is about where that hour goes missing, what Zodiux does with the two clock hours each year that have no honest answer, and which parts of your chart you should still believe when the time is bad or absent.

One hour of error, ranked by what it breaks

Everything in a chart moves. It is the rates that differ — by four orders of magnitude between the fastest and slowest rows below — and that spread is the whole story of birth-time accuracy.

Each figure is the change in geocentric apparent ecliptic longitude over one hour, sampled quarterly from 1960 to 2026 using the same Astronomy Engine build that runs in your browser when you press Read my chart. Negative values are retrograde motion — a body appearing to track backwards along the zodiac.

Read the first row against the last. In the hour it takes an unrecorded birth time to drift, Pluto moves about a hundredth of a degree and the rising sign moves half a sign. Those are not the same kind of uncertainty and they do not deserve the same confidence.

Why the rising sign is the only fragile placement

Your Sun sign is a fact about where Earth is in its orbit. Your rising sign is a fact about which way the ground under you was facing. The Earth turns once in a sidereal day, so the zodiac degree climbing over the eastern horizon advances at an average of 15.04 degrees per hour — some 366 times faster than the Sun. That single ratio is why one number on the form carries almost all the risk.

The arithmetic that turns a moment into an ascendant is set out line by line, with a worked example and its error budget, on how Zodiux calculates a birth chart. What that page does not do is show how badly the average misleads.

15 degrees an hour is an average, and the average lies

The ascendant does not advance smoothly. The ecliptic is tilted against the horizon, so some signs rise almost flat and take hours to clear while others rise steeply and are gone in under an hour. The effect strengthens with latitude, and it is entirely predictable: the twelve durations pair up symmetrically and sum to exactly one sidereal day, 1,436 minutes, which is how we know the measurement is right.

At the equator the ascendant never leaves the band 13.8 to 16.4 degrees per hour, and no sign takes less than 111 minutes to rise or more than 129. At London's latitude the instantaneous rate ranges from 10.5 to 36.0 degrees per hour: Pisces and Aries are over in under an hour, Virgo and Libra take nearly three. So "the rising sign changes every two hours" is a serviceable rule of thumb and a bad basis for confidence. A 40-minute error is harmless if you were born with Libra rising in London and decisive if you were born with Pisces rising there.

This is also the practical answer to "another site gives me a different rising sign". If you were born at a high latitude with a short-ascension sign on the horizon, two calculators an hour apart on the time zone will disagree, and neither is doing the astronomy wrong.

Sign by sign, with each boundary crossing bisected to the second in the shipped ascendant function:

The two clock hours each year that have no clean answer

You type a local time; the astronomy needs a UTC instant. Zodiux gets from one to the other by reading your birth city's IANA zone name and asking the browser's own copy of the time-zone rules what offset applied on that date — not today's offset. That is one lookup, and it happens once.

One lookup is enough for 8,758 hours of the year. It is not enough for the other two. When clocks go back, one local hour happens twice; when they go forward, one local hour never happens at all. Zodiux resolves both without asking you, and it is worth knowing which way, because both are one-hour decisions and one-hour decisions are rising-sign decisions.

The split is not a policy. It falls out of the one-pass lookup: the offset is read at a provisional instant made by reading your wall clock as though it were already UTC, and whether that provisional instant lands before or after the clock change depends on the sign of the zone's offset. Zones west of Greenwich take the earlier reading; zones on or east of it take the later one.

We ran the shipped conversion against every one-hour clock change in 1975, 2000 and 2020 in all 326 zones the city file uses. The repeated hour resolved to the earlier instant in 131 of 324 cases and to the later one in the other 193. The missing hour moved forward in 131 of 323 cases and backward in 192 — never by anything other than exactly one hour. One zone breaks the west/east rule: Atlantic/Azores, whose summer offset is exactly UTC+0.

What the time-zone database does model

The folklore is that calculators get wartime and energy-crisis births wrong. In the data a modern runtime actually ships, they do not — those periods are modelled, and Zodiux inherits the modelling for free because it carries no time-zone rules of its own. Every offset below was read back out of the standard date-formatting API, sampled at 15 January of each year so that summer time is never mistaken for a change in the standard offset.

Where the time-zone data stops

Two limits are worth stating plainly, because between them they cover most of the births this page cannot help.

The first limit is a date: 1970

The IANA project's stated scope is timestamps since 1970. Data before that is best-effort, and since 2021 the default distribution has gone further: zones whose clocks have agreed since 1970 are merged into one rule set, with the older differences moved to a backzone file most platforms do not compile in.

That is observable, not theoretical. In the runtime we tested this page against, these zone names return identical offsets at every quarterly sample from 1900 to 1969: Europe/Oslo, Europe/Copenhagen, Europe/Stockholm and Europe/Berlin; Europe/Amsterdam and Europe/Brussels; Atlantic/Reykjavik and Africa/Abidjan; America/Nassau and America/Toronto; Europe/Monaco and Europe/Paris. Some of those places genuinely did keep the same clocks. Others did not — Iceland ran an hour behind Greenwich until 1968, and the Netherlands kept its own offset about twenty minutes ahead of it until 1937 — and those differences now live only in the file most browsers never load.

Counted across the whole shipped city file: 121 of the 326 zone names return an identical pre-1970 offset history to at least one other zone, in 29 indistinguishable groups. (Narrow the comparison to 1930 onward and it is 200 in 45 groups — the further back you look, the more the histories separate.) If you were born before 1970 in a place since folded into a larger neighbour, your offset is an inference — and a twenty-minute inference is already a 1-in-6 shot at the wrong rising sign.

The second is harder: your town was not the zone's capital

A zone name is a representative city, not a municipal register. In the United States between 1945 and 1966 there was no federal daylight-saving law and towns chose for themselves — sometimes differently from the county next door, and differently from year to year. The database does carry some of that: the city file uses six separate zone names for Indiana, Kentucky and North Dakota alone, and in June 1965 they were not all on the same clock as each other, let alone as New York.

But six names cannot cover a decision made town by town. Zodiux gives you the rule for the zone your birth city is filed under. Whether the specific hospital clock followed that rule on that day is not something any calculator can know.

The clock that wrote your birth certificate

A perfect time zone cannot repair a bad source number, and the source is usually worse than people assume.

A recorded birth time is a bureaucratic artefact. Someone read a clock — or reconstructed it later — and wrote it on a form. Several things routinely go wrong at that step, and none of them are visible on the document afterwards.

Ordinary ways a certificate time drifts

• The form is not the birth. In a busy delivery the paperwork is completed after the fact, sometimes considerably after, and the number written is the one recalled or the one on the nearest clock.
• Clocks in the room were not synchronised. Wall clock, monitor and wristwatch can disagree by minutes — and on a clock-change night, by an hour.
• "The moment of birth" is not one moment. Delivery of the head and complete delivery of the body can be minutes apart, and practice differs on which one is written down.
• Recorded times cluster on round numbers. This is the one claim on this page we did not measure ourselves: digit preference, or heaping on the hour, the half hour and the five-minute marks, is a long-standing finding in the vital-statistics literature, not something Zodiux computes.
• Many jurisdictions record no time at all, or record it only on the long-form copy held by the registrar rather than the short certificate you were sent.

This is not an argument for despair. It is an argument for calibration. A certificate time you trust to five minutes leaves the rising sign right about 24 times in 25. A time recalled by a relative decades later, to the nearest hour, is a coin flip — a different object entirely, and it should not be given the same weight just because the calculator prints both to the same number of decimal places.

No birth time: what the chart prints and what it withholds

When the time field is left blank, Zodiux uses local noon on your birth date in your birth city. Noon is chosen because it is the midpoint of the day, which halves the worst-case error on every body compared with assuming midnight.

What happens next is deliberately uneven, because the underlying certainty is uneven. The percentages are measured by comparing a noon chart against the true chart for every 13th minute of the day, across 1950, 1975 and 2000.

Two rows deserve more than a cell. The Moon: the chart shows a single Moon sign even when the time is unknown, and it is simply wrong 11% of the time. The moon sign calculator handles the same input better — with the time blank it checks the sign at 00:01 and at 23:59 local and, if they differ, names both possibilities rather than choosing. If you have no birth time and your Moon matters to you, run that tool as well as the chart. What a moon sign is has the day-by-day detail.

Aspects are the row most people misjudge. A no-time chart looks stable because the planets barely move, but the printed list is a cut at the eight tightest, and the Moon forms about 2.5 of the roughly 12 aspects in an average chart while crossing 6–7 degrees over half a day. Aspects between the slow bodies do hold up; the list as a whole does not. Read the non-lunar rows and discount the lunar ones — what aspects are explains what each one claims to mean.

How much to trust each placement, given what you actually know

Find the line that matches your evidence. Every figure is the measured flip rate from the top of this page, applied to the size of the error you are probably carrying.

From strongest evidence to weakest

1. A certificate time to the minute, and the chart does not put your ascendant in the first or last degree of a sign. Trust everything printed. The remaining uncertainty is narrower than the lines in the wheel.
2. A certificate time you would defend to about five minutes. Rising sign right about 24 times in 25. Everything else unaffected.
3. A certificate time, but the ascendant lands in the first or last degree of a sign. Treat the rising sign as two candidates and read both: one degree of ascendant is worth roughly four minutes of clock, which is less than most certificates are good for.
4. A time recalled by a parent, to the nearest quarter hour. Rising sign right about 7 times in 8. Everything else unaffected.
5. A time recalled to the nearest hour, or a clock-change night. Rising sign is a coin flip. The Moon sign survives 98 times in 100; Sun and the outer planets are certain.
6. "Sometime in the morning", or nothing at all. Do not use a rising sign. The Sun, Mercury, Venus and Mars come out right about 99 times in 100 from the date and city alone — Mercury is the weakest at 99.0%, because it can cover more than a degree in a day — and the outer planets are effectively fixed. The noon Moon is right about 9 times in 10. That is most of a chart, and Zodiux shows all of it — it just will not show you a rising sign.

Rectification: the thing Zodiux does not do

There is a traditional practice called rectification: working backwards from events in someone's life to a birth time that would have produced a chart matching them. Some astrologers offer it as a paid service.

Zodiux does not do it and will not add it. The reason is method, not squeamishness. Rectification takes an unknown quantity and fixes it using the very interpretations it will then be used to justify, so the output cannot be checked against anything independent of the assumption that produced it. Everything else on this site is falsifiable: take any position we print, feed the same instant to any other ephemeris, and you get the same number. A rectified birth time is not that kind of claim, so it does not belong on a page that publishes its own error bars.

If your birth time is genuinely unrecoverable, the honest options are a noon chart with the rising sign omitted — which is what you get here by default — or a request to the registrar or hospital that holds the long-form record.