Before deciding whether astrology is accurate, turn the dial below to 0°. The outer ring says Aries; the inner one lands in Pisces. Neither answer is a calculation error — they answer different questions. We will start with that mismatch, then work backwards through ancient calendars and forwards to experiments that can actually fail.
What is the difference between astronomy and astrology?
Astronomy studies the positions, motions and physical nature of celestial bodies, and its predictions — the date of the next lunar eclipse, say — can be checked against observation. Astrology reads the positions of planets and zodiac signs as influences on human character or fate; no physical mechanism for that is known, and in controlled tests it has not performed better than chance. In antiquity both were written on the same clay tablets. They part ways over whether a claim can be tested.
Has astrology been tested scientifically?
Yes, carefully. In 1985 Carlson agreed an experimental design with astrologers in advance and asked them to match birth charts to one of three personality profiles; over a hundred judgements later, they did no better than the one-in-three chance rate. In 1948 Forer showed that a vague description written to fit anyone leaves most people feeling it "got them right". Cultural meaning can be studied, but a real-world prediction has to pass tests like these.
Hold the verdict and follow the evidence
Turn the dial before taking a side. It shows that a zodiac sign, a constellation, and ecliptic longitude are three different things. Then Sirius, calendars, and research data can tell us how far each kind of claim actually goes.
Sirius really was a good marker
Sirius vanishes for about seventy days each year, then reappears one morning just before sunrise. Ancient Egypt treated that day as the start of the year.
Around the same time, the Nile began to rise. Year after year, the two matched.
But the water came from monsoon rain on the Ethiopian highlands, not from the star. They happened together because both follow the same annual cycle.
A heliacal rising depends on latitude, so somewhere else it falls on another date — and the Nile takes no notice. A real cause stays a cause when you move.
A calendar that slowly walked out of season
The Egyptian civil year was 365 days: twelve thirty-day months plus five at the end, with no leap day.
So it lost a day every four years, a month every hundred and twenty, and needed 1,460 years to come back round. Any given date drifted through every season.
In 238 BCE someone proposed adding a day every fourth year. It did not stick. Real farming still ran on the river and the stars.
Over the same centuries, scribes in Mesopotamia were also recording the sky, except what they recorded was numbers.
One tablet does not mean one kind of evidence
Babylonian scribes tracked the Moon and planets for centuries and found that eclipses repeat about every eighteen years. That period alone let them predict next year's eclipse, with no theory required.
The same people also wrote omens: which sky event meant what for the king and the harvest. Both kinds of writing sit on the same tablets.
The prediction half turned out to work. The omen half was never tested. Sharing a tablet does not make them stand or fall together.
They also cut the ecliptic into twelve equal thirty-degree parts, for a practical reason: writing 'so many degrees of such a sign' beats reciting star names. Today's twelve signs come from there.
Now look back at the dial you just turned
The twelve signs split 360 degrees into equal thirty-degree slices, counted from the equinox.
Constellations are sky regions whose borders were only drawn in 1930, and they vary a lot: the ecliptic covers 44 degrees of Virgo, under 7 of Scorpius, and passes through Ophiuchus on the way.
The dial above turns. Across the whole 360 degrees, the two systems agree on a name for only 44 of them.
Then there is precession: the equinox slides west by a degree every seventy-two years, so zero degrees of Aries now falls inside Pisces. Nobody miscounted; the two systems were tied to different things from the start.
‘That sounds like me’ is not a hit rate
A classroom experiment in 1948: students rated 'their' personality analysis for accuracy and averaged 4.26 out of 5. Every one of them had been handed the same text.
A description that is broad and flattering enough will feel accurate to almost anybody.
So judging accuracy means setting up a control group, blinding the people involved, and agreeing in advance on what result would count as a failure.
So 'it felt accurate' cannot serve as evidence. Feelings matter, but they cannot measure a hit rate.
Ask a precise question and a study can answer it
Carlson's 1985 double-blind test in Nature was designed with astrologers in advance. Asked to match a chart to one of three personality profiles, they scored no better than chance over a hundred-plus attempts.
In 2006 another group went through records for 4,462 middle-aged men and 11,448 young adults and found no relationship between birth date and personality or intelligence.
What those results reject is the claims that were tested. Astrology remains worth studying as history and culture; the moment it claims to describe a person or an event accurately, it has to be measured with the same ruler as everything else that predicts.
Four questions are yours to reuse: what exactly does this predict? What result would count as wrong? Where are the comparison group and the blinding? Can somebody else repeat it under the same rules?
Try it first: one angle, two names
Press ‘0°: equinox’ and read the sign and constellation in the centre. Then try 195° and 256°. The outer ring uses fixed 30° signs; the inner ring follows the actual constellation boundaries.
After a full turn, look at the thin green ring. Only 44° out of 360° — about 12% — has the same name in both systems. That is not a scorecard; it is a reminder to ask which coordinate system is being used.
0°: the equinox
The signs are counted from here, so 0° opens Aries. On the equinox the Sun actually sits in Pisces — the two systems were never pinned to each other.
195°: the widest constellation
The ecliptic runs nearly 44° through Virgo, more than six times its stretch through Scorpius. A sign is always 30°; a constellation is whatever its boundary says.
244°: the narrowest constellation
The ecliptic crosses barely 7° of Scorpius, while the sign of the same name gets a full 30°. Same word, more than four times the difference in sky.
256°: the constellation with no sign
The ecliptic spends 18° in Ophiuchus, and the twelve signs have no slot for it. Nobody miscounted; the two systems answer different questions.
Model limit: this is a 1° teaching diagram on the J2000 reference ecliptic, not a dated ephemeris, birthday converter, natal chart, or personal sign tool. Do not read sub-degree precision near a boundary.
Start with Sirius: a clock, not a cause
Sirius disappears for about seventy days each year, lost in daylight too close to the Sun. Then, on one particular morning, it shows up low in the east a few minutes before sunrise and is gone again as the sky brightens. That return is its heliacal rising.
Ancient Egypt marked the day and treated it as the opening of the year. Around the same stretch of the season, the Nile began to rise. The two lined up, and kept lining up for three thousand years.
The water did not come from Sirius. Most of the flood came down the Blue Nile and the Atbara, fed by monsoon rain over the Ethiopian highlands from June to September. The rain fell 1,500 km upstream and took weeks to arrive. Nothing connects that star to that rain; they happened together because both hang on the same orbit of the Earth.
There is a ready-made test in this. A heliacal rising depends on latitude, so the same Sirius rises on a different date somewhere else — and the Nile takes no notice. A real cause is still a cause when you move; two things that merely coincide come apart.
The Egyptians got nothing wrong. The marker worked for three thousand years because it really was accurate enough. It was a clock the whole time.
Ancient scribes put calculations and omens on the same tablets
The Egyptian civil calendar ran twelve thirty-day months plus five days at the end: 365 days, no leap day.
Sirius comes back around in a little more than 365.25 days, so the calendar slipped a day every four years, a month every hundred and twenty. A given civil date drifted slowly through every season, and needed about 1,460 years to come back round. Farming could not run on it; that ran on the river and the stars.
In 238 BCE the Canopus Decree proposed adding one day every fourth year. It was not carried through, and the civil calendar kept drifting for another two centuries.
Over the same centuries, scribes in Mesopotamia were doing something else. The MUL.APIN tablets, compiled around 1000 BCE, hold star lists, the dates stars rise heliacally, rules for intercalation, and the ratio of day length to night across the year.
Later Goal-Year tablets computed the future straight from periods. Lunar eclipses repeat roughly every 223 synodic months — about eighteen years — and Jupiter returns to the same place in about twelve. Shift centuries of old records forward by one period and next year's prediction falls out, with no theory of why anything moves that way.
The same scribes wrote omens: which month an eclipse fell in, where a planet stopped, and what that meant for the king and the harvest. Those lines sit on the same tablets, in the same hands.
The twelve equal divisions of the ecliptic show up in the texts around the fifth century BCE, and they show up as a convenience — writing the Moon's place as 'so many degrees of such a sign' beats reciting star names. The twelve signs astrology uses today and the ecliptic longitude astronomy calculates with come from that one tool.
Their eclipse predictions turned out to work. The omen half was never tested. Both were written on the same tablet, and that fact is not evidence for either one.
A tablet noting where a planet 'stood still' is describing what we now call a station: the moment a planet switches from direct to retrograde motion, or back. Why it reverses at all is what the retrograde lesson draws from real orbital data — along with how much of the year Mercury actually spends retrograde.
Turn the dial: a sign is not a constellation
Ecliptic longitude is an angle, counted from the equinox, 360 degrees around the sky. The Sun, Moon, and planets all travel near that line, so one number places any of them.
The twelve signs cut those 360 degrees into equal thirty-degree slices, the first one starting at the equinox. That is the tropical zodiac, and most Western astrology uses it.
Constellations are a different thing. In 1922 the International Astronomical Union fixed the names of 88 of them; in 1930 the Belgian astronomer Delporte drew their borders along lines of right ascension and declination for 1875. Since then a constellation has been an administrative district of the sky rather than a join-the-dots picture. The ecliptic passes through thirteen of them, Ophiuchus included.
The districts vary wildly in width. The ecliptic runs nearly 44 degrees through Virgo and under 7 through Scorpius. That is what the inner ring of the dial above shows: out of a full 360 degrees, the two systems produce the same name for only 44.
Then there is precession. The Earth's axis swings like a slow top, once in something over 25,000 years, and the equinox slides west by a degree every seventy-two years. Zero degrees of the sign Aries now falls inside the constellation Pisces.
The site also has a lesson called perihelion precession. The name is easy to confuse with the precession above, but it is a different motion: the long axis of a planet's orbit turning, not Earth's spin axis — the equinox does not move because of it.
This gets written up as 'your sign has been wrong all along'. But the tropical zodiac is tied to the seasons by definition and not to the stars, and the people using it do not need it to line up with a constellation. Whether the signs match the constellations is not really the question. The question is the next step: what that position is then said to mean.
If a claim can never be wrong, it cannot be tested
In 1948 the psychologist Bertram Forer gave thirty-nine students a personality test and handed back 'their' analysis a week later. The students rated it for accuracy and averaged 4.26 out of 5. Everyone had received the same text, assembled from a newsstand astrology book.
Every personality reading has to clear that hurdle first. A description broad enough, flattering enough, and close enough to ordinary human experience will feel accurate to almost anyone. Judging whether a method works means keeping that effect outside the door.
The method is not complicated. Say in advance what is predicted, how it is scored, and what result would count as a failure; run a control group that gets a fake reading; keep both the people judging and the people being judged blind to which goes with which; then let somebody else repeat the whole thing under the same rules.
None of this is aimed at astrology in particular. Medicines, weather models, and stock-picking methods all face the same hurdle the moment they claim to predict. The only difference is whether anyone has actually run the test.
What exactly does the claim predict?
What result would count as wrong?
Where are the comparison group and blinding?
Can others repeat it with the same rules?
The tests were run — and the questions were specific
In 1985 Shawn Carlson published such a test in Nature. The design was negotiated in advance with astrological organisations, who agreed it was a fair test of what they do. That agreement is rare, and it is what makes the paper worth reading.
Around two dozen practising astrologers, nominated by their own organisations, took part. Subjects filled in the California Psychological Inventory; each astrologer received one natal chart and three CPI profiles and picked which profile belonged to the chart. Over a hundred judgements later, the hit rate was indistinguishable from one in three. The astrologers had predicted they would reach one in two.
The second half ran the other way: subjects chose their own reading out of three interpretations of their chart. That did no better than guessing either.
In 2006 Hartmann, Reuter, and Nyborg took a different route and went through existing large datasets: 4,462 middle-aged men in one, 11,448 young adults in the other. They looked for relationships between birth date and measured personality or general intelligence, and found none; the Sun-sign associations they tested came up empty too.
These are solid negative results, and their reach is exactly that wide: these instruments, these samples, this approach. The accurate wording is that these tests provided no support — not that two papers closed the file on every astrological tradition.
Cultural meaning can be studied; predictions must be tested
Ptolemy wrote the Almagest, the astronomy textbook for the next fourteen hundred years. He also wrote the Tetrabiblos, the astrology textbook for the next fourteen hundred years. Same man, same geometry.
Kepler's day job was Imperial Mathematician, but the salary was often in arrears, so he moonlighted as an astrologer, casting charts for paying clients — the one he drew for Wallenstein in 1608 is still famous. In the same years, with one hand casting horoscopes, he wrote the Astronomia Nova with the other, replacing circular planetary orbits with ellipses. He also complained in print about how his fellow practitioners went about it: criticism from inside the trade.
That history rewards reading, and what it teaches is how people hang meaning on the sky: how a symbol forms, who uses it, and where it sits in a society. Those questions can be researched, and the answers can be right or wrong.
Astronomy asks a different set: where it is, how it moves, what it is made of, and why it behaves that way.
Only one thing actually separates the two — whether a sentence is about meaning, or is a claim about the world that could be shown to be false. The first belongs to history and culture. The second has to queue up with every other method that claims to predict.
First identify what kind of sky claim you are reading
Layer
How it is established
What it answers
Observation / coordinate
Repeat a measurement from a shared reference
Where an object is in the sky
Classification convention
Divide 360° equally by agreement
How to label a longitude sector
Astronomical sky region
Apply formal IAU boundaries
Which background region contains a sightline
Personality / event interpretation
Make a falsifiable prediction and blind-test it
Whether accuracy exceeds chance
FAQ
Sirius really did mark the Nile season. Why is that not evidence for astrology?
Ancient Egypt marked the day and treated it as the opening of the year. Around the same stretch of the season, the Nile began to rise. The two lined up, and kept lining up for three thousand years. The water did not come from Sirius. Most of the flood came down the Blue Nile and the Atbara, fed by monsoon rain over the Ethiopian highlands from June to September. The rain fell 1,500 km upstream and took weeks to arrive. Nothing connects that star to that rain; they happened together because both hang on the same orbit of the Earth. There is a ready-made test in this. A heliacal rising depends on latitude, so the same Sirius rises on a different date somewhere else — and the Nile takes no notice. A real cause is still a cause when you move; two things that merely coincide come apart.
Why do zodiac signs and astronomical constellations not line up?
The twelve signs cut those 360 degrees into equal thirty-degree slices, the first one starting at the equinox. That is the tropical zodiac, and most Western astrology uses it. Constellations are a different thing. In 1922 the International Astronomical Union fixed the names of 88 of them; in 1930 the Belgian astronomer Delporte drew their borders along lines of right ascension and declination for 1875. Since then a constellation has been an administrative district of the sky rather than a join-the-dots picture. The ecliptic passes through thirteen of them, Ophiuchus included. The districts vary wildly in width. The ecliptic runs nearly 44 degrees through Virgo and under 7 through Scorpius. That is what the inner ring of the dial above shows: out of a full 360 degrees, the two systems produce the same name for only 44. This gets written up as 'your sign has been wrong all along'. But the tropical zodiac is tied to the seasons by definition and not to the stars, and the people using it do not need it to line up with a constellation. Whether the signs match the constellations is not really the question. The question is the next step: what that position is then said to mean.
Has astrology actually been tested scientifically?
In 1985 Shawn Carlson published such a test in Nature. The design was negotiated in advance with astrological organisations, who agreed it was a fair test of what they do. That agreement is rare, and it is what makes the paper worth reading. Around two dozen practising astrologers, nominated by their own organisations, took part. Subjects filled in the California Psychological Inventory; each astrologer received one natal chart and three CPI profiles and picked which profile belonged to the chart. Over a hundred judgements later, the hit rate was indistinguishable from one in three. The astrologers had predicted they would reach one in two. The second half ran the other way: subjects chose their own reading out of three interpretations of their chart. That did no better than guessing either.
This lesson covers clearly testable parts of Western zodiacal astrology. External links are references, not runtime providers. The lab uses a J2000 reference ecliptic and whole-degree approximate IAU boundaries; it does not compute ancient heliacal risings for a place and era, and it does not create natal charts.