Van Gogh’s Starry Night matches a real predawn sky from July 1889, not June

The crescent Moon, the blazing morning star and the knot of stars at the centre of Van Gogh’s The Starry Night line up with a real sky, according to a new study, and the sky they match is not the one art historians assumed. A team in South Korea reports that the arrangement fits the dawn over Saint-Rémy-de-Provence about a month after the painting’s accepted date, when a thin waning Moon, Venus and the V-shaped Hyades cluster rose together in the east. If the match holds, one of the most reproduced night skies in Western art was painted from observation, not invented.

That reverses a long-running assumption. For decades the swirling sky above the village was read as religious vision, memory or pure imagination, with Venus the only object most scholars accepted as real. The new analysis, published in Scientific Reports, argues that the Moon and the central star group are real too, and that the painter kept their positions with surprising accuracy while turning everything around them into motion.

The study was led by Elijah Jeong-hyun Kim, chief executive of the South Korean space-science company SLLAB and a doctoral candidate at Kyung Hee University’s School of Space Research, with colleagues Suna Kang, Jungwoo Song and Ho Jin. Their best-fitting date is a dawn around 22 July 1889, while Van Gogh was a patient at the Saint-Paul-de-Mausole asylum.

“No two skies are the same. Something always changes. I believe Vincent noticed those changes and put them into the painting,” Kim told the Associated Press.

How they did it

The team did not start with The Starry Night. They first tested their procedure on other night paintings. Adam Elsheimer’s The Flight into Egypt failed the test: its Big Dipper sits at an angle to the Milky Way that no real sky can produce. Jean-François Millet’s Starry Night passed, with Orion’s belt, Rigel and Sirius placed close to their true positions. The method was then calibrated on two of Van Gogh’s own night scenes from Arles, Starry Night over the Rhône and Café Terrace at Night, whose painting spots still exist and can be visited.

The key step was recovering how much of the sky the painter actually framed. Van Gogh built the landscape of The Starry Night from an earlier drawing of the hills seen from the asylum. By surveying the ridge south-west of the asylum and overlaying the drawing on field photographs, the team found a narrow field of view, equivalent to a 134 mm telephoto lens, or about 18 degrees across the diagonal.

Applied to simulated skies over Saint-Rémy, that frame is very selective. On 21 July the Moon and Venus sat almost 30 degrees apart, too far to fit. On 22 July the gap had closed to 18.6 degrees, with the Hyades at the centre and the Moon only 28 percent lit. By 23 and 24 July the two had drawn too close together to match the composition.

To test the star group itself, the researchers used Procrustes analysis, a geometric method that scales and rotates one set of points to measure how closely it matches another shape. Five stars in the painting were matched to the five skeleton stars of the Hyades. The fit was close, if somewhat looser than for the two Arles paintings. Out of 120 possible ways of pairing the five painted stars with the five real ones, the anatomically correct pairing was the single best fit. The tilt of the painted V, 33.13 degrees to the horizon, also falls inside the 32.6 to 35.1 degrees the real cluster showed before dawn that night.

A pen copy that Van Gogh made of the painting matched the Hyades more closely than the oil, about as well as the Arles works, which the authors read as a sign that he was working from the real arrangement and not copying his own invention. A previous examination of the canvas had already found that he left gaps for the stars before painting the blue sky, so their placement was decided early.

Why the June date never fit

The standard date comes from a letter of 18 June 1889, in which Van Gogh told his brother Theo about “a new study of a starry sky.” In 1984 the art historian Albert Boime proposed the dawn of 19 June and read the central stars as Aries. Two years later the astronomer Charles Whitney reached a different answer and called the sky “a celestial composite.” Their incompatible conclusions helped push the observational reading to the margins.

The Moon was always the weak point. On 19 June 1889 it was about 63 percent lit, a lopsided Moon still well past half, nothing like the sharp crescent on the canvas. José Luis Aragón, a physicist at the National Autonomous University of Mexico who has studied the painting and was not involved in the work, called that mismatch “a real puzzle” in comments to the AP.

The team adds two more threads. Van Gogh later grouped The Starry Night with Olive Trees, a picture he tied to cicadas and midday heat, and weather records for Avignon show only four days at or above 30 °C in June 1889 against eighteen in July. The authors also re-date his second seizure, usually placed in mid-July, to late July: he wrote that it struck while he was painting in the fields “on a windy day,” and a strong mistral blew across the region from 27 to 29 July.

What it doesn’t settle

The study does not say when Van Gogh started or finished the canvas, or whether he reworked it. He painted in his studio by day, and the authors themselves describe the sky as a synthesis of several adjacent nights centred on 22 July. Aragón put it plainly: “Still, I would see the July date as an intriguing hypothesis rather than a settled answer.” He added that the sky “is likely a composite of many nights as the authors themselves acknowledge.”

The geometric fit is not proof on its own. The team shows that other star patterns, such as Aries or parts of Lyra and Auriga, can produce similar or even better shape scores, and they are ruled out only on other grounds: where they sit in the sky, which star is Venus, and how few points they offer. The weather records from 1889 are patchy and partly qualitative, and no medical record from the asylum fixes the start of the seizure.

The landscape is also a studio reconstruction, built from an earlier drawing and not painted on the spot. What the study claims is narrower and harder to dismiss: if the Moon, Venus and the Hyades match a specific dawn this closely, the burden shifts to anyone who says the sky was imagined. “What I find most convincing is the broader message,” Aragón said. “The painting is not pure fantasy. Van Gogh watched the real sky very closely.”

Common questions about The Starry Night sky

When did Van Gogh paint The Starry Night?

Art historians date it to June 1889, from a letter Van Gogh wrote to his brother Theo on 18 June. The new study argues that the sky it shows matches the dawn of around 22 July 1889, which suggests that the work continued into July.

Is the sky in The Starry Night real?

According to this study, the main elements are. The crescent Moon, Venus and the central group of stars match the eastern sky over Saint-Rémy-de-Provence before dawn in late July 1889. The swirls, the halos and the size of the stars are artistic exaggeration.

What is the bright star next to the cypress in The Starry Night?

It is Venus, the morning star. Van Gogh wrote to Theo that he had watched the countryside from his window before sunrise “with nothing but the morning star, which looked very big.”

What are the Hyades?

The Hyades are the nearest open star cluster to the Sun, about 150 light-years away in the constellation Taurus. Their brightest stars form a V shape near the red giant Aldebaran, and the study identifies that V as the star group at the centre of the painting.

Where can you see The Starry Night?

The painting hangs at the Museum of Modern Art in New York. Kim told the AP that he has never seen it in person: “Now that the paper is out, I really have to go and see ‘The Starry Night’ in person, don’t I?”

The authors say their method can now be applied to other historical night paintings, including many whose skies have never been checked against astronomy. “I am confident that July 22 is the better date,” Kim said. “I will bet on this date.”

Reference: Kim et al., “The actual sky of Van Gogh’s Starry Night: a multidisciplinary reconstruction,” Scientific Reports, 2026. DOI: 10.1038/s41598-026-73025-5

Tags: astronomy, Marc Anthony, MarShawn Lloyd, Vincent van Gogh, Venus, art history