What Did He Actually See?

How an Astrophysicist Spent Thirty Years Solving the Mystery of Van Gogh's Starry Skies

In any well-designed bookshop around the world, Van Gogh merchandise occupies a permanent corner. Those swirling stars appear on mugs, phone cases, tote bags. The image has become a kind of cultural shorthand — for the suffering genius, for art that permits madness, or for nothing at all, just something that looks good.

But nobody has ever seriously asked: which night sky did he paint?

Everyone tells you Van Gogh painted with his soul.

Those swirls, those stars, those burning cypresses — they have been interpreted as the delirium of genius, the visual residue of a mental breakdown, the scream of a man abandoned by his era inside an asylum. This has been the standard narrative of art criticism for more than a hundred years, and no one has meaningfully questioned it.

Then, in 1995, a French astrophysicist picked up a thin book in a secondhand shop, and started asking a question of almost embarrassing simplicity:

Were those stars real stars?

If so, which ones? On what night? At what hour?

These questions, and the thirty years required to pursue them, changed everything we can know about those paintings.

A Scientist's Eye

Jean-Pierre Luminet is Provençal by birth, a specialist in black holes and cosmology. He also draws and paints — charcoal, gouache, oil — and at some point discovered Van Gogh.

He says he was "immediately struck by the way this painter had translated his visions of the nocturnal sky." But his astonishment was different from an art critic's. Where critics saw emotion, he saw a question that could be precisely verified.

The secondhand book was a French translation of an article published in the United States in 1984 by Albert Boime, professor of art history at UCLA, under the title Van Gogh's Starry Night: A History of Matter and a Matter of History. Boime had attempted to use astronomical reconstruction to determine which real sky Van Gogh had painted in The Starry Night. The approach excited Luminet — but he quickly found serious flaws in Boime's conclusions.

He sat down with his astronomical software, Voyager, and began calculating for himself.

That calculation took thirty years. Along the way, he made multiple visits to Saint-Rémy to verify the orientation of the asylum windows and the surrounding topography in person. He established contact with descendants of Van Gogh's treating physicians, gaining access to medical records. He read through all 902 letters in the complete digital edition — which the Van Gogh Museum only published in full in 2018. In 2023, he assembled his findings into a book: Les Nuits Étoilées de Vincent Van Gogh (Éditions Seghers, Paris).

Why He Wanted to Paint the Night Sky

Before unravelling the mystery, one must understand why it was difficult in the first place.

When Van Gogh arrived in Arles in February 1888, he already had the idea of painting a starry sky — but he kept putting it off. On June 19 he wrote to his painter friend Émile Bernard: "When will I ever do the starry sky, that picture that has always preoccupied me?… Alas, alas, the most beautiful paintings are those one dreams of while smoking a pipe in one's bed, but which one never makes." [628]

Luminet points out that this was not ordinary self-doubt. He was intimidé par le défi technique — intimidated by the technical challenge. He did not know how to paint night at night.

The standard practice for painters of the time was to sketch by day and add color back in the studio. The resulting nocturnes were literary imaginations of night, not the visual facts of night itself.

Van Gogh refused this. He wanted to paint outdoors, at night, facing the actual sky. But this meant identifying pigments in darkness and judging tonal relationships under artificial light. He admitted as much in a letter: "It is quite true that in the dark I may mistake a blue for a green, a blue-lilac for a pink-lilac, since it is hard to distinguish the quality of a tone clearly. But it is the only way to escape from the conventional black night with a poor, pallid and whitish light — when in reality even a simple candle already gives us the richest yellows and oranges." [678]

He accepted the trade-off, took up the technical problem, and began painting.

He also had one advantage others lacked. His eye was, in Luminet's phrase, un œil exercé — a trained eye. He could perceive color gradations in the night sky that no one else noticed. "It often seems to me," he wrote, "that the night is still more richly coloured than the day, having hues of the most intense violets, blues and greens. When you pay attention you will see that certain stars are lemon-yellow, others pink or a green, blue and forget-me-not brilliance." [678]

This was not metaphor. It was a description of an astronomical fact — stars do exhibit a range of color temperatures, from red to blue-white, depending on their surface temperature. Van Gogh perceived this with the naked eye, with no knowledge of astrophysics, and judged it worth painting.

Café Terrace at Night: The First Verification

In September 1888, Van Gogh finally acted. The first painting to genuinely capture a night sky was Café Terrace at Night.

Luminet used Stellarium to reconstruct the actual sky visible from the Place des Hommes (today Place du Forum) in Arles, looking due south, between September 9 and 14, 1888. He compared the star positions in the painting. The stars correspond to those of the constellation Aquarius. The positional relationships match the real sky closely. Brightness is greatly exaggerated — but the relative positions are faithfully preserved.

This small verification established the foundation of Luminet's entire study: Van Gogh was pursuing positional accuracy, even while exaggerating color and intensity — what he himself called being a coloriste arbitraire, an arbitrary colorist.

Starry Night Over the Rhône: A Century Before Photoshop

A week later, he painted Starry Night Over the Rhône. This canvas now hangs in the Musée d'Orsay in Paris.

The painting contains a secret that every art critic who has written about it for the past 130 years has entirely failed to notice. Luminet noticed, because he is an astronomer.

At the center of the canvas, you can recognize the seven stars of Ursa Major — the Big Dipper. But Ursa Major is a circumpolar constellation: it circles the North Pole and can only be seen in the northern sky. It never appears above the southern horizon.

Yet the terrestrial landscape in the lower half of the painting — the Rhône, the city skyline, the bridge — is the view looking southwest from the Arles riverbank.

You cannot see Ursa Major when looking southwest. These two views cannot coexist in a single field of vision.

Art critics missed this entirely, because none of them thought to check the position of the constellation.

Luminet calculated Van Gogh's exact easel position (43° 40' 56" N, 4° 37' 48" E) and reconstructed his stance at the riverbank: on his left, the city's gas lamps and the southwestern view of the Rhône; on his right, looking north, Ursa Major rotating in the night sky. He compressed both directions into a single frame.

"He produced one landscape in the lower part of the canvas and another in the upper part, gathering all the space that surrounded him into a single frame… A century before the advent of Photoshop."

But this is not even the painting's most striking feature.

Luminet and photographer Raymond Martinez made a further discovery: Van Gogh had non sans malice — not without malice — positioned the stars in alignment with the gas lamp reflections. The orange columns of gaslight from the street lamps extend downward into the Rhône, and the stars above are positioned to align precisely with those lamps. The viewer is presented with an illusion: the reflections on the water appear to belong to both the lamps and the stars simultaneously, light running unbroken from river surface into sky.

Not without malice. This is the most precise phrase in Luminet's entire book for describing Van Gogh. It says nothing about genius, nothing about madness. It describes a craftsman's deliberate cunning — a visual trap laid on canvas, waiting for someone with sharp enough eyes to stumble into it two centuries later.

The couple in the foreground, for their part, stands on a patch of water with no reflections at all — because directly above that spot is the bridge, and gas lamps do not reach water beneath a bridge, and no bright stars happen to sit in the sky directly overhead. This detail, once you have understood the engineering precision of the whole composition, feels inevitable.

The Starry Night (Saint-Rémy): 4:40 in the Morning

In May 1889, Van Gogh voluntarily admitted himself to the Saint-Paul-de-Mausole asylum in Saint-Rémy. The Starry Night was completed over the following weeks. It now hangs at the Museum of Modern Art in New York.

This painting has attracted more astronomical speculation than any other. Boime argued that the great swirling form in the center was Van Gogh's reproduction of an image of the Whirlpool Nebula (M51, in Canes Venatici), which had been drawn by the astronomer Lord Rosse in 1845 and widely reproduced in popular astronomy literature — including in Camille Flammarion's famous Astronomie populaire of 1881.

Luminet dismissed the hypothesis quickly.

First layer of evidence: in 902 letters, astronomy is never mentioned.

Van Gogh reported every reading source to Theo without exception — Zola, Michelet, Hugo, Maupassant, Dickens, George Eliot — down to the most minor detail. Yet in all 902 letters, Flammarion's name appears exactly zero times. The word "astronomy" appears exactly zero times. If he had genuinely been inspired by astronomical imagery, silence on this point alone would be inexplicable.

Luminet's final judgment: "For him, painting the sky and the stars in their realistic configurations represented a pictorial challenge and obeyed solely a purely aesthetic imperative."

Second layer of evidence: the swirl is how he painted clouds.

Luminet compared the great swirling form in The Starry Night with the cloud forms in several canvases Van Gogh completed during the same period — June to July 1889: Wheatfield After a Storm, Wheatfield with Cypress, Olive Trees with the Alpilles in the Background. The brushstroke grammar is identical. This is not a nebula. It is his established method for rendering moving air — serpentine, coiling, advancing strokes that took shape in Arles and reached their peak in Saint-Rémy.

So what are the real astronomical elements in the painting?

In 1995, Luminet calculated the sky visible from Saint-Paul-de-Mausole looking east, for every night before the painting's likely completion date. He found the closest match at one specific moment:

May 25, 1889, at 4:40 AM local time.

This configuration lasted approximately fifteen minutes. Venus (the morning star) hung low on the horizon, correctly positioned to the right of the cypress. The crescent moon was illuminated from below in the correct proportion — the unmistakable signature of a pre-dawn moon. Hamal and Sheratan, the two brightest stars of Aries, flanked the cypress tip on either side. Two bright stars of Pisces sat between the moon and Venus.

Van Gogh suffered frequent insomnia during this period. He wrote to Theo: "This morning I saw the countryside from my window a long time before sunrise with nothing but the morning star, which looked very big." [777]

Luminet's reconstruction: on the night of May 25, during an episode of insomnia, Van Gogh looked through the iron bars of his window and saw this configuration. He made a precise sketch. Two or three weeks later — once Theo had sent new canvas and paint — he began the oil painting in the asylum's studio. As of July 2, his letters show he was still reworking it.

He spent approximately six weeks completing this supposedly spontaneous masterpiece.

Road with Cypress and Star: The Canvas as Mirror

In the final weeks before leaving Saint-Rémy, Van Gogh completed another nocturne — Road with Cypress and Star, sometimes called Country Road in Provence by Night. A tall cypress stands beside a road, a crescent moon and two bright stars hang in a deep blue sky, and a horse-drawn carriage passes with two figures walking nearby.

Luminet's analysis of this painting reveals another startling feature of Van Gogh's compositional method.

The crescent moon in the painting is illuminated from the right — meaning the sun has just set below the left horizon, consistent with a twilight scene at dusk. But in astronomical reality, if the sun sets on the left, the crescent moon should appear in the right side of the sky. The moon in the painting faces the wrong way.

Through computer reconstruction, Luminet confirmed that on April 20, 1890, just after sunset, Venus, Mercury, and the crescent moon formed an unusually close alignment — three celestial bodies in a near-vertical line, a rare configuration occurring only once every few years. Van Gogh observed this event, committed it to memory, and returned to his studio to paint.

But he faced a familiar compositional problem: the cypress as central vertical axis divides the sky into two halves. Placing the three bodies in their correct positions would unbalance the composition.

Luminet's finding: Van Gogh took the left half of the sky and mirror-flipped it, reassembling it as the right half — preserving the relative positions of the moon and the two planets while satisfying the symmetrical demands of the overall composition. The result of this operation is an astronomical error in the direction of the moon's illumination — and that error is precisely the fingerprint of Van Gogh's compositional engineering.

Without an astronomical background, this detail cannot be found. Before Luminet found it, the painting had long been regarded as a relatively casual nocturnal sketch.

Luminet's conclusion: "Van Gogh obeyed only the formal requirements of the canvas — he worked from a real framework, but he knew how to depart from it when his artistic intuition demanded." This is a precise footnote to Delacroix's dictum: "A painting is a machine whose every system is intelligible to a trained eye."

Four Elements in Composition

Together with psychiatrist Philippe André, Luminet traced the complete sources of The Starry Night and reached a final judgment: the painting is composed of four precisely assembled elements.

First, real star positions. Venus, the crescent moon, and the stars of Aries as observed through iron bars on the morning of May 25 at 4:40 AM — noted in an immediate sketch, with a precision that defies casual inspiration.

Second, his cloud-painting grammar. The great swirling form is not a nebula. It is his established method for rendering moving air — consistent with canvases from the same weeks.

Third, a Dutch village from memory. The village with its needle-sharp steeple is not Saint-Rémy. The asylum window faces east; Saint-Rémy and its church lie to the north, entirely invisible from that vantage. Luminet compared the church spire in the painting with those in works by Van Goyen and Van Ruysdael — the shape belongs to the Dutch landscape of his childhood, a projection of his native land placed in the space where the asylum's wheatfield would actually have been.

Fourth, a real Provençal cypress. The compositional axis dividing the canvas; simultaneously, in the original Greek sense, a symbol of the bond between life and death, earth and sky, finite and infinite — "a symbol of night even in full daylight."

"The Starry Night expresses an admirable triple conjunction: precise observation, imagination, and memory."

The Studio Engineer

This is the most important claim in Luminet's book — and the one most capable of dismantling the received narrative of Van Gogh:

Van Gogh was not driven by instinct to reproduce what he saw as fast as he had seen it. He was a builder.

On July 1, 1888, he wrote to Theo: "Know that I am in the midst of a complicated calculation, from which a succession of canvases rapidly result — painted quickly, but worked out beforehand over a long period. So that when they say it is done too quickly, you can reply that they looked too quickly." [635]

He said this exactly once in 902 letters.

He never described his compositional method to Theo. He reported the names of pigments — ultramarine, cobalt blue, chrome yellow, lead white. He gave Theo the general meaning of each canvas, described the scene before him. But how he organized all of this into a painting, how he executed those compositional montages — not a word. Luminet calls this silence: C'est sa cuisine — that is his kitchen. His private business.

In Arles, he averaged one canvas per day. In the fifty-three weeks at Saint-Rémy, he completed 143 oil paintings and more than 100 drawings. In the final seventy days at Auvers, he produced over eighty canvases. This pace demands massive pre-calculation. It cannot be sustained by impulse alone.

Art criticism read his silence and his speed as "the instinct of genius." Luminet reads them in precisely the opposite way: precisely because the compositional engineering was so rigorous, it required long preparation in advance; and precisely because he did not want to reveal the process, he kept silent with Theo.

The Thirty-Year Paradox

Luminet's research has the structure of a beautiful paradox.

He spent thirty years of astronomy proving that Van Gogh had no interest in astronomical science.

The star positions in the paintings come from the real sky, but only because he was a painter of extraordinary perceptual precision — able to see colors in the night sky that no one else perceived, and to translate that perception into the physical facts of pigment. This required the training of a painter, not the knowledge of an astronomer. He painted stars not to make scientific records but because those stars presented him with a pictorial challenge that needed to be solved.

Luminet opens his book with a passage by the philosopher Gaston Bachelard about Van Gogh's color — one that now reads as a footnote to the entire thirty-year investigation: "A yellow by Van Gogh is like an alchemical gold, a gold harvested like solar honey. It is never simply the gold of wheat, of flame, or of a straw chair: it is a gold forever individualized by the endless dreams of genius. It no longer belongs to the world, but is the treasure of one man, the heart of one man, the elementary truth found in the contemplation of an entire life."

Bachelard was a philosopher's observation. Luminet is an astronomer's verification. They arrived at the same conclusion from opposite directions: Van Gogh's color is the product of a perceptual capacity developed through lifelong training, not the accidental gift of descending genius.

This conclusion makes both art criticism and astronomical research simultaneously central to the story and somehow beside the point. Luminet admits that it took him thirty years to properly understand that Van Gogh's act of peindre le ciel — painting the sky — ultimately obeyed one command only: the aesthetic command.

Van Gogh wrote a sentence in September 1888, one that gets quoted often and understood rarely: "I have a terrible need — shall I say the word — of religion, and so I go outside at night to paint the stars."

Luminet reads this sentence differently from every art critic who has ever cited it. He reads it not as emotional confession but as a technical declaration: Van Gogh needed a certain kind of anchor that transcended the ordinary, and the only way he could actively manufacture that anchor was to go outdoors at night, face the real sky, and record it in paint.

Religion was a perceptual need. Painting was the perceptual technique. He fulfilled both in the same gesture.

Luminet's book ends with a quiet image: he imagines Van Gogh in that magic interval between dawn and dusk, walking alone through the Provençal fields with an easel on his back, his head full of celestial swirls. Those swirls, 130 years after his death, were decoded by an astrophysicist with software and precise calculation — and found to be not madness, but a rigorous, long-calculated reassembly of the night.

Source Jean-Pierre Luminet, Les Nuits Étoilées de Vincent Van Gogh, Éditions Seghers, Paris, 2023. ISBN 978-2-232-14620-6. All letter citations correspond to the chronological numbering established by the Van Gogh Museum and the Huygens Institute, Amsterdam; complete letters available at vangoghletters.org.