The Starry Night, stroke by stroke

You are Vincent van Gogh, born in 1853 in Zundert in the Netherlands. You worked for the art dealers Goupil before deciding, in 1880, to teach yourself to be an artist, with your brother Theo paying your way.

Your schooling was short: a winter at the Antwerp Academy in 1885–86, then Fernand Cormon's studio in Paris from 1886. Paris gave you Japanese prints, which you collected, and painters such as Émile Bernard, Toulouse-Lautrec and the Neo-Impressionists.

In February 1888 you moved south to Arles. Paul Gauguin stayed with you from October to December and argued for painting from imagination; you had always painted from the motif in front of you. By 1889 you were looking back to Delacroix, Millet and Daubigny, and to Monticelli's thick paint.

The idea is a year old. In April 1888 you wrote of wanting to paint a starry night with cypresses. In September 1888 you painted a starry sky over the Rhône at Arles on the spot, at night, under a gas lamp, and a café terrace at night without any black.

This time you can't paint on the spot. You have been in the asylum since 8 May. You may go out only by day and with an escort, and you are never out at night. One morning before sunrise you saw from your window only the morning star, and it looked very big. The editors of your letters identify it as Venus.

Decide these before you touch the canvas

From memory

You will compose rather than copy: a view built by day in the studio from memory and from your own recent pictures. Gauguin urged painting from imagination; in June Theo warns you about its “mysterious regions”.

The parts

Hills from your painting of the view from your window, a cypress from your painting of a green wheat field, and the village from a pencil sketch drawn on the slope above the town.

The sky

A night sky with the morning star, a crescent moon and big haloed stars. Whether you are recording a real sky or inventing one is still argued (see chapter VIII).

Why

For consolation, for what you call style, and in competition with the pictures by Gauguin and Bernard that you can only imagine.

At Saint-Paul-de-Mausole you have two rooms. Your bedroom's barred window looks east over a walled square of wheat to the hills, and you watch the sun come up over it. A second room is yours to work in.

You paint The Starry Night by day, in that second room. The story that you painted it outdoors at night belongs to the Rhône picture of the year before.

Early in June you painted the view from your window. Its line of hills will reappear in the new picture, but the walled field will not, and the village could not be seen from your window at all.

Around the end of June you make a reed-pen drawing after the painting and send it to Theo on 2 July with other drawings. It adds a star and smoke from a chimney.

You hold the canvas back from a shipment in September and send it on 28 September, probably rolled. By then you group it with two other canvases as exaggerations of arrangement.

Theo is wary. In October he writes that the search for style dulls true feeling, and you reply defending a drawing that is “plus mâle et plus volontaire”, more manly and more deliberate. In November you tell Émile Bernard that you let yourself paint stars too big, and call it another failure.

Astronomers have long tried to read your sky, and only one point is safe: the bright star near the cypress is probably Venus, the morning star of your letter. Older readings found the constellation Aries or a composite of several nights.

In October 2026 a team from SLLAB Inc. and Kyung Hee University published a new reconstruction in Scientific Reports. They simulated the sky over Saint-Rémy in planetarium software checked against NASA's JPL ephemerides, fitted the painted stars to real star positions, surveyed the site in 2019, and checked weather records and your letters. They tested the method first on Millet's Starry Night and on your own Rhône and café terrace pictures.

Their argument runs like this. In mid-June 1889 the moon was gibbous, about 63% lit, which does not match your crescent. At dawn on about 22 July 1889, a waning crescent about 28% lit, Venus and the V-shaped Hyades star cluster all fit the frame of the view, and five central painted stars match the Hyades, more closely in the drawing than in the painting. They suggest you watched a real sky, stylised it, and finished the canvas in July.

Hold it lightly. It is one new study, and it conflicts with the letters edition, which places the drawing among those sent on 2 July. Other astronomers called exact dating fragile and suggested the crescent may be artistic licence. The authors themselves say they cannot tell when the painting was begun or whether it was reworked, that a good shape fit is not proof of which stars are shown, and that the sky may combine several nights. Three of them run the company that may profit from the work.

Some other claims have not held up: a 2024 paper said the swirls follow the physics of real turbulence, and in 2025 other scientists disputed it.

Less than you might expect. No colour change has been documented in this painting, and no treatment report has been published, so its history of varnish, lining and retouching is not public.

• After Theo, it belonged to his widow, Jo van Gogh-Bonger. A Rotterdam collector, Georgette van Stolk, later shaded it with a curtain in summer, and it seems to have escaped damage from sun and damp.
• The dealer Paul Rosenberg sold it to MoMA in 1941, in exchange for two Cézannes and a Toulouse-Lautrec. It was the first Van Gogh in a New York museum, on view from 30 September 1941.
• In 2019 MoMA's conservators took it out of its frame to examine it, with a conservation grant from Bank of America.
• Your drawing after it was taken from storage in 1945 and is held in Russia, out of public view.