Ayanamsa: the 1.4° that can move a Moon across a boundary

Every Indian matching report begins by deciding which nakshatra your Moon was in, and that decision rests on a number different astronomers have set differently for over a century. Everything downstream depends on the answer: the nadi group, the yoni animal, the gana, the counted distance between two stars, the whole 36 points. It looks like the most factual line on the page. It is the line with a convention buried inside it.

What is the ayanamsa actually subtracting?

The accumulated drift between the tropical zodiac, which is tied to the equinox, and the sidereal zodiac, which is tied to the fixed stars. Both are ways of dividing the same circle. They do not agree about where the circle starts.

Western astrology uses the tropical frame. Indian astrology uses the sidereal one. The two coincided roughly seventeen centuries ago and have been pulling apart ever since, at about a degree every seventy-two years, because the Earth’s axis wobbles slowly and the equinox slides backwards through the stars. The size of that accumulated gap is the ayanamsa, and it is what a program subtracts from a tropical longitude to get the sidereal longitude an Indian chart is drawn from.

The drift itself is not controversial in the slightest. Its rate has been measured for a very long time; the Surya Siddhanta is working with the phenomenon well over a thousand years ago, and modern values agree with each other to a precision that would have astonished its authors. Nobody argues about the rate.

What is unsettled is the anchor. Where, exactly, does the sidereal zodiac begin? Which star marks it, at what longitude, at what epoch? That is not a question an instrument answers, because the stars themselves carry no markings. It is a question about what a tradition intended, and it has been answered several different ways by several serious people.

Why do published ayanamsa values disagree at all?

Because each one anchors the sidereal zodiac to a different reference, and each of those choices is defensible. No arithmetic mistake has been made by anybody. The values differ because the starting assumptions differ.

The Lahiri value, adopted by the Calendar Reform Committee and used in the Indian national ephemeris, places Chitra at 180°. That is a clean, stateable rule, it produces a value that can be computed for any date, and it carries official standing in India. It is what the Positional Astronomy Centre works with, and it is what nearly every widely used piece of matching software runs by default.

B. V. Raman used a different derivation and arrived at a value roughly a degree and a half smaller. His Manual of Hindu Astrology sets out his reasoning, and the southern reading that descends from him formally specifies his value rather than Lahiri’s. The Krishnamurti value sits a few arcminutes from Lahiri. Fagan–Bradley, used mostly outside India, sits close to a degree the other side.

So there are at least four values in circulation, spanning something like two and a half degrees end to end, all of them arrived at honestly. This is not a scandal and it is not a fringe dispute. It is the ordinary condition of a question that has no observational answer.

What do the classical texts say about where the zodiac begins?

Less than the argument needs them to. This is the part that surprises people who expect the dispute to be settled by seniority, and it is the reason the dispute has survived so long.

The classical siddhantas are working astronomy. They set out methods for computing positions, they know about precession, and they give values for quantities their authors could check. What they do not do is state, in a form that survives translation into a modern coordinate frame, the exact longitude of a named star at a named epoch in a way that pins the sidereal origin to within a few arcminutes. The instruments of the period were not capable of that precision, so it would have been an odd thing to write down.

Chitra — Spica — is the usual anchor because a reading of the texts places it at or near the midpoint of the circle, and it is a bright, unmistakable star sitting close to the ecliptic. That makes it an excellent practical choice. It does not make it a stated one. Other scholars have anchored to Revati at 0°, or worked backwards from a dated observation, or from the position of a solstice at a reconstructed epoch. Each of those lines of reasoning is a serious piece of scholarship and they do not converge on a single figure.

Which produces an outcome nobody planned. The most consequential single number in Indian astrology — the one that decides which star every chart is read from — was fixed in the twentieth century by committee, on scholarly grounds, from texts that do not quite answer the question. The committee did a careful job and its answer has the best claim of any. It is still a twentieth-century decision about a first-millennium intention, and a reading that follows a different scholar is not being unorthodox.

How large is a degree and a half, in the units that matter?

About a tenth of a nakshatra, and closer to half of a pada. Stated as a fraction of a full circle it sounds like nothing. Measured against the objects it moves, it is not nothing at all.

A nakshatra spans 13°20′ — the circle divided by 27. The pada, the quarter of a nakshatra that matching actually reads for several of its rules, spans 3°20′. Set a degree and a half against those two numbers and the scale becomes obvious. It is a tenth of a birth star and something like four tenths of a quarter.

Two astrologers can be looking at the same Moon, at the same instant, and reading two different nakshatras from it.

The pada figure is the one people underrate. A great deal of matching reads the nakshatra alone, but the exhaustive population this site measures over is the 11,664 ordered pairings of nakshatra-and-quarter, precisely because the quarter carries real weight in several traditions. Something that moves the pada boundary four tenths of the way across a pada is a substantial input, not a rounding term.

There is a second comparison that lands harder for most readers. A degree and a half of frame shift is roughly what you get from moving a birth time by six minutes, for the ascendant, or by about three hours for the Moon. Nobody would accept a report that had been computed from a birth time three hours wrong. A great many people accept, without ever being told, a report computed on a frame that a different documented school would have set a degree and a half elsewhere. The two are not the same kind of error — the frame is a choice and the wrong time is a mistake — but they are the same size, and only one of them gets checked.

How often does a Moon sit close enough to a boundary to matter?

Roughly a tenth of the time, which sounds small until you count how many families that is. The Moon moves about 13° a day, so it crosses a nakshatra boundary roughly once a day, and it spends about a tenth of each of those days within a degree and a half of one.

The arithmetic follows directly. A degree and a half is about a tenth of 13°20′, and it applies at both ends of the interval, so a Moon has something like a one-in-ten chance of sitting in a band where the birth star depends on the ayanamsa. Put another way: a chart born within roughly two and a half hours of a nakshatra crossing is a chart where the setting is doing the deciding.

For the other nine charts in ten, none of this matters even slightly. Every published value gives the same birth star, the report would be identical under any of them, and the ayanamsa question can be dismissed entirely. That is worth saying clearly, because the honest version of this page is not that all readings are uncertain. Most readings are not. A specific, identifiable minority are, and the report can tell you which you are — the positions tool prints the Moon’s degree within its nakshatra, which is the whole test.

What changes downstream when the birth star moves?

Not one line of the report. All of it. The birth star is not one of the inputs to a matching calculation. It is the input from which the others are read.

Nadi is read from the nakshatra, in three groups of nine. Yoni is read from it. Gana is read from it. Tara is counted from it. Bhakoot and graha maitri are read from the rashi, which can also change if the shift happens to cross a sign boundary as well. Move the star and the whole assessment is recomputed from a different starting point. The result is not a small correction to the previous total; it is a different total, built from different components, which can perfectly well be higher.

Nadi is the sharpest case, because it carries the largest single block of points — 8 of the 36 — and it pays all or nothing. Two people in the same nadi group score zero on it. Measured over every possible birth-star pairing rather than a sample, that happens on 33.3% of them, which makes it one of the commonest single reasons a match is refused outright. Whether a given couple falls in that share can turn on a boundary a fraction of a degree away. There is a full account of nadi elsewhere; what matters here is only that it is decided by which of three groups the star falls in, and the star is decided by the ayanamsa.

The neighbouring nakshatra is rarely in the same nadi group, because the groups cycle in threes. So a one-star shift across a boundary changes the nadi assignment about two-thirds of the time. That is the mechanism by which a setting nobody printed turns into the sentence that stopped a marriage.

Worth noticing which direction that runs in, though. A star that moves can move into agreement as easily as out of it, and the shift is as likely to clear a dosha as to raise one. People reading about this for the first time tend to hear it as evidence that their refusal was wrong. It is not evidence of anything about a particular chart. It is a statement about how firmly the input was established, and a chart sitting in the sensitive band is one where the refusal and the acceptance are both less settled than they look.

Is the uncertainty in the arithmetic or in the convention?

Entirely in the convention, and the gap between the two is enormous. These get confused constantly, and separating them is the single most useful thing on this page.

The planetary positions themselves are not in doubt. This engine’s output is checked against NASA’s JPL DE440 ephemeris across two centuries of dates; over 2,000 sampled epochs from 1900 to 2100, the worst disagreement in any position it computes is 0.44 arcseconds. An arcsecond is a 3600th of a degree.

Hold those two numbers next to each other. The gap between two published ayanamsa values is more than a degree. The gap between our arithmetic and NASA’s is under half an arcsecond. The convention is thousands of times wider than the computation, and the same is true of any competently written program — the underlying DE440 solution is public and everybody serious builds on it or on something equivalent.

Which is the useful way to hold this. Nobody’s computer is imprecise here. No future improvement in instruments or software will narrow the disagreement by any amount at all, because the disagreement is about a choice made before the computing starts. That is a strange and slightly liberating thing to know. It means the question is not technical and cannot be settled by whoever has the better software.

Does the ayanamsa move the manglik verdict as well?

It can, through the ascendant rather than the Moon, and that route is much less discussed. Mangal dosha is decided by which house Mars occupies, and houses are numbered from the ascendant.

The ascendant moves through roughly a degree every four minutes of clock time, so shifting the frame by a degree and a half is equivalent to moving the birth time by about six minutes. Most of the time that changes nothing. If the ascendant happens to be sitting near a sign boundary, the whole house numbering rotates by one, and a Mars that was in the 12th is now in the 1st — which, on most house lists, is the difference between clean and flagged.

The published sample makes the point without any hypotheticals. The bride’s ascendant in that case moves from Tula to Vrishchika on a shift of just 3 minutes in the recorded birth time. A chart that sensitive to the clock is equally sensitive to the frame, and the two effects are indistinguishable on the printout. Whether a chart is in that condition is knowable in advance, which is the point of the four-minute birth-time test.

How does this affect the divergence figure published here?

It makes it a floor. This is the part we should be straight about, because it cuts against our own headline number rather than for it.

When we ran every one of the 11,664 possible pairings under five readings — the whole population of nakshatra-pada combinations rather than a sample of it — and found that 56.4% trigger at least one tested difference or veto across its checks, that census held the ayanamsa fixed at Lahiri for all five. Including for the school that formally specifies Raman’s value.

Letting the ayanamsa vary as well would move the number up, not down, because it adds a source of divergence rather than removing one. We publish the figure with the axis held fixed because holding it fixed is the conservative choice and because a figure that is a floor is easier to defend than one assembled from every effect we could find. It is worth knowing that the real spread across genuinely different implementations is wider than what we quote.

There is a second, quieter consequence. Because almost all software runs Lahiri regardless of which school it claims, a nominally southern reading in circulation is usually a southern rule set on a northern frame. That hybrid appears in no text either, and it is the reading a great many families actually received — which is the same pattern as the merged manglik house list, and for the same reason. The settings a program chooses on your behalf are almost all like this.

What do we use, and what can you check on a report you already have?

We compute on Lahiri, every report says so on the page where the positions are printed, and the check you can run on any other report takes about a minute. Alongside the ayanamsa we print the birth time we used and the coordinates we resolved for the birthplace, because those are the other two inputs that can silently move a star.

Where a Moon or an ascendant sits close enough to a boundary that a plausible change in the inputs would move it, the report says so rather than printing a birth star as though it were beyond question. A chart that is genuinely sensitive should not be presented with the same confidence as one that is not, and the difference between the two is computable.

To run the check yourself: find the page of your existing report that lists planetary longitudes, and read the Moon’s position as a degree within its nakshatra. More than about a degree and a half clear of both ends and the ayanamsa question is closed for your chart; you can stop thinking about it and go and argue about something that is actually in dispute. Inside that band, ask which ayanamsa was set and what the birth star becomes under the other one. The nakshatra finder will give you the second answer directly.

What none of this licenses is a conclusion about a person. A shifted birth star changes which rules a tradition applies and what those rules conclude. It does not make a statement about a marriage, and neither do we — the list of things this site declines to claim is set out in what we do not predict, and the ayanamsa is a good illustration of why that list has to exist. A setting this consequential, this invisible and this unresolvable is not a foundation anyone should be making predictions from.