Rahu Is a Lunar Node: How Hindu Astronomy Predicted Eclipses While Keeping the Demon

Open a printed panchanga to today's date and you will find Rahu listed with a longitude, given to the arcminute, moving backwards through the zodiac at roughly three minutes of arc a day. Ketu sits exactly one hundred and eighty degrees away. In that column neither has a face, a chariot or a grudge. They are the two points where the Moon's orbital path crosses the Sun's apparent path, and an eclipse can only occur when a new or full Moon happens near one of them.

That is the answer to the jibe that Hindus believe a demon eats the Sun. The Puranas do say a severed head pursues the Sun and Moon and swallows them. The siddhantas say a solar eclipse is the Moon covering the Sun, and a lunar eclipse the Earth's shadow falling across the Moon. Both statements belong to the tradition, one word carries both, and the almanac that predicts the eclipse to the minute still prints the demon's name at the top of the column.

The oldest layer: a hymn about the Sun being hidden

Rigveda 5.40, from the Atri family collection, describes something covering the Sun. In Griffith's rendering, the Asura's descendant Svarbhanu pierces Surya through and through with darkness, and all creatures look like one who is bewildered and does not know where he is standing (verse 5). Indra smites the magic that had spread itself beneath the sky, and Atri finds Surya concealed in gloom (verse 6). Atri then establishes the eye of Surya in heaven again and makes the magic arts vanish (verse 8), and the hymn closes by saying the Atris found the Sun and that none besides had power to do it (verse 9).

The commentarial tradition reads Svarbhanu as Rahu under an older name. Whether or not that was in the poet's mind, the hymn already has the shape of the whole later complex: something obscures the light, the obscuring is hostile and temporary, and ritual knowledge held by a particular lineage restores it.

The hymn uses no technical vocabulary for nodes, syzygy or shadow, and it does not say the obscuring body is the Moon. Attempts to pin it to a specific historical eclipse (some propose fourth-millennium BCE dates) rest on unsettled assumptions about the Rigveda's chronology, and they do not agree with one another.

The severed head

The story most people know comes from the churning of the ocean. In the eighth canto of the Bhagavata Purana, Vishnu takes the Mohini form and serves the amrita so that the devas receive it. Rahu covers himself in the dress of a deva and takes his place in the line. The Sun and the Moon, seated nearby and already his enemies, give him away. Hari's disc, sharp as a razor, takes off his head before the nectar has passed below the neck, so the body cannot survive and the head, which the nectar has reached, does not die (8.9.24 to 25). The verse after that is the one that matters for the almanac: Brahma accepts the immortal head as one of the grahas, and it attacks the Sun and Moon on the nights of full Moon and dark Moon (8.9.26). The severed trunk is Ketu in the later reckoning; the Bhagavata itself says only that the body died.

The grudge follows from the informers. Rahu chases the two who exposed him, and because there is no body below the neck the swallowed light comes out again: total, then over. Full Moon and dark Moon, the nights the Purana names as his opportunity, are precisely the two phases at which an eclipse is possible at all.

What is less often noticed is how much apparatus the Puranas attach to him. The Vishnu Purana puts Rahu in the same catalogue of celestial chariots as the Sun and the Moon: eight black horses draw his dusky car, and on the parvas he directs his course from the Sun to the Moon and back again from the Moon to the Sun (2.12, in Wilson's translation). The parvas are the new and full Moon. That is a statement about periodic motion tied to syzygy, sitting inside a Purana.

The Bhagavata goes further in its cosmological canto. It puts Rahu ten thousand yojanas below the Sun and says he moves like one of the stars (5.24.1), gives his extent as thirty thousand yojanas (5.24.2), and says he cannot stay in front of the Sun or Moon longer than a muhurta, forty-eight minutes (5.24.3). A distance, a size, a maximum duration. None of the numbers agree with the siddhantas, and a Rahu below the Sun belongs to a cosmos in which the Sun is nearer than the Moon, which the astronomers rejected outright. But the form of the claim is the form of a parameter rather than of a fable.

What a node actually is

The Moon's orbit is tilted about 5.1 degrees to the plane of the Earth's orbit, so the Moon spends half of each circuit above that plane and half below, crossing it at exactly two points. The line joining those crossings is the line of nodes. Where the Moon passes from south to north is the ascending node, which the Indian tradition calls Rahu; the opposite one is the descending node, Ketu.

This is why eclipses are rare. At new Moon the Moon lies between us and the Sun, at full Moon the Earth lies between Sun and Moon, but usually the Moon rides several degrees too high or too low and the shadow misses. Only when that alignment happens near a node do the three fall close enough to a straight line for a shadow to land.

The numbers are tight. A solar eclipse becomes possible when new Moon falls within roughly 15 to 18 degrees of a node, the spread coming from the changing distances of Sun and Moon, and it can be total or annular only inside about 12 degrees. For a partial umbral lunar eclipse the limit is nearer 9 to 12 degrees, for a total one nearer 4 to 6. The line of nodes slides backwards, completing a circuit in about 18.6 years, which is why Rahu takes around eighteen months to cross one rashi. And because the draconic month, node to node, is 27.212 days while the synodic month is 29.531 days, the two cycles drift against each other and throw up eclipse seasons about 173 days apart. Eclipses come in clusters, then nothing for half a year.

Rahu is not something that arrives. It is a place where a line crosses a plane, and nothing sitting there causes anything. Being near the node is the geometric condition under which Sun, Moon and Earth are capable of lining up at all, which is exactly what the Puranas encoded as the demon's opportunity.

Aryabhata says it plainly

In 499 CE, in the Golapada of the Aryabhatiya, Aryabhata puts the mechanism in a single line, with no argument and no polemic around it.

चन्द्रो जलमर्कोऽग्निर्मृद्भूश्छायापि या तमस्तद्धि

छादयति शशी सूर्यं शशिनं महती च भूच्छाया

"The Moon is water, the Sun fire, the Earth soil, and its shadow darkness. The Moon covers the Sun, and the great shadow of the Earth covers the Moon."

Aryabhatiya, Golapada 37

W. E. Clark's 1930 edition heads that verse with nothing more decorative than "Moon Causes Eclipse of Sun; Shadow of Earth Causes Eclipse of Moon". Aryabhata does not stop to refute anyone. He says what does the covering, and the verses after it go to work: when eclipses occur, the length of the Earth's shadow, its diameter at the Moon's distance, how much of the disc is obscured. The refutation, such as it is, is the arithmetic. If you can compute the shadow and predict what a person in a particular place will see, you have said everything that needs saying about what causes the darkness.

Varahamihira argues the case out loud

Varahamihira, writing in the sixth century, is less restrained. The fifth chapter of his Brihat Samhita is called Rahucara, the course of Rahu, and its opening verses report the views he has inherited: that Rahu is an asura who tasted the ambrosia, that he is serpent-shaped, that he is a bodiless darkness. Then he takes them apart.

The truth, he says, is that in her own eclipse the Moon enters the shadow of the Earth, and in the Sun's eclipse she enters the solar disc (5.8). That shadow keeps neither the same direction nor the same length from one night to the next, any more than a tree's does, because the Sun moves (5.9). When the Moon, whose course is always from west to east, stands opposite the Sun and swerves neither much north nor much south, she enters it (5.10). Moving from the west, she hides the solar disc from below like a cloud, and the solar eclipse therefore varies from country to country according to how much of the covered disc is visible (5.11). Then the closing blow: what eclipses the Moon is bigger than the Moon, and what eclipses the Sun is smaller than the Sun (5.12). One swallowing body cannot be both.

Notice how little of the chapter that occupies. Rahucara runs to ninety-eight verses, most of them on what eclipses portend for kings, armies and crops. Varahamihira dismantles the demon in a dozen lines, then works on under the demon's name.

One word, two registers

This is what a working panchanga does today. Rahu and Ketu appear among the nine grahas, each with a computed longitude, exactly opposite each other, normally retrograde. They are classed as chhaya grahas, shadow planets, an unusually precise piece of terminology: they are not bodies, and what they mark is the place where a shadow becomes possible. Almanac makers still argue about whether to publish the mean node, which regresses at a steady rate, or the true node, which oscillates around it under the Sun's pull, drifting away by as much as a degree and a half and occasionally turning direct. That is enough to shift Rahu into a different nakshatra pada, so the argument has consequences. But it is about which computation better matches the sky, not about who Rahu is.

The Sanskrit word for eclipse is grahana, from the root grah, to seize. Graha, the ordinary word for planet, is from the same root: a seizer. The astronomers who explained that nothing seizes the Sun wrote every line of their technical prose in a vocabulary built on the idea that something does. Nobody appears to have found this a problem.

On the day itself the two registers do different work. The computed timings tell a household when the contact begins, when totality falls, when it ends. The story tells them why they bathe afterwards, why they do not eat during it, why the mantra is what it is. They are not two competing theories of the same fact, and it takes a particular kind of reading to insist that they must be.

Where the fight actually was

It would be too smooth to say the two accounts sat side by side for a thousand years without friction. The friction is on the page, in Sanskrit, and knowing where it sat changes the picture.

Lalla, writing in the eighth or early ninth century, gave a whole chapter of his Shishyadhivrddhidatantra to mithyajnana, false knowledge, and spent it demolishing exactly these ideas. If Rahu has a body, or is a disc, or a head, and if he is always moving, why should he swallow the Moon only at a distance of six signs from the Sun? If an artful demon is the cause, how is it that an eclipse can be determined by calculation? In a solar eclipse people in different parts of the Earth see different portions covered, and some see nothing at all. Later in the chapter he turns on the claim that the Earth is flat like a mirror: if it were, why does water falling from the sky not stand still, instead of running off at speed in one direction? (In Bina Chatterjee's edition this is chapter twenty, the flat-earth arguments at verses 34 to 37.)

Bhaskara II repeats the mirror simile in the Goladhyaya of the Siddhanta Shiromani (1150) and answers it: if the glorious Earth is flat like the belly of a mirror, why is the Sun, revolving far above it, not always visible to human beings as it is to the gods? In his own commentary, the Vasanabhashya, he names the source outright. It is described in the Puranas, he writes, that the Earth is flat like the belly of a mirror. He does not say where, and no one since has found the passage. You do not name your opponent's scripture unless somebody has been pressing you with it.

Christopher Minkowski called the long-running settlement the "standard accommodation", substantially standardised without ever being quite stable or uniform. The astronomers absorbed the Puranic material that could be made to fit, placing Mount Meru at the north pole, and rejected what could not, above all the ordering that puts the Sun closer to us than the Moon. First articulated in Lalla, it was carried by Shripati in the eleventh century and Bhaskara II in the twelfth.

It did not hold forever. Around 1500 the astronomer Jnanaraja, in his Siddhantasundara, reversed the polarity. He never rejects a cosmographical idea taken from the Puranas. He defends them against Bhaskara II by name, endorses the view that the Earth needs the support of divine beings to keep its place, and redefines what "up" means so that a Puranic description can stand. Toke Knudsen, whose study of the episode sets out most of this, points to what Jnanaraja does not do: he says nothing at all about the claim that the Sun is nearer to us than the Moon. He cannot absorb it, because if the Sun sits below the Moon the siddhantic model breaks down and eclipses cannot be computed. Rather than expose the rift, he goes quiet.

That is the honest shape of it. The dispute was real, and it was mostly about cosmography rather than the eclipse mechanism, on which the astronomers were never seriously contested. Their account predicted; the other did not. Nobody had a motive to abandon the name, the story or the rites, which had never been in the business of predicting anything.

When someone laughs at the swallowed Sun

The mockery assumes a tradition can hold only one description of a thing at a time. The reply that fails is "we never believed that," because the Puranas do say it and Hindus still say grahana and still say Rahu grasta. The other reply that fails is "our ancient texts already contained modern astronomy," because they did not. Aryabhata worked it out, Varahamihira had to argue it, Lalla had to write a chapter against people who would not have it, and the Bhagavata's numbers for Rahu are wrong.

The accurate reply is narrower and much harder to answer. By 499 CE a named Indian astronomer had written that eclipses are shadows and had given the geometry to compute them. Within a century another was refuting the demon by argument, and by the ninth a third had given a chapter to it. For the thousand years after that, the almanacs predicted eclipses from the position of a point on the ecliptic that carries the demon's name, and the families who used them went on telling the story of the severed head.

Open the panchanga on the morning of the next eclipse and both are on the page. The name in the first column is the demon's. The number in the second is a longitude computed from the Moon's orbit. The household reading them will know when the contact begins, and will bathe when it is over.