On March 11, 1890, the German Chemical Society filled Berlin's city hall to honor August Kekulé. The occasion, remembered as the Benzolfest, marked 25 years since his first paper proposing that benzene, a liquid of six carbon and six hydrogen atoms, was built as a closed ring. Kekulé was 60 and in poor health. When his turn came to reply, he told the audience where the idea had come from.
He had been dozing in front of the fire, he said, and saw chains of atoms twisting like snakes. One of them took hold of its own tail. He woke with a start and spent the rest of the night working out the consequences.
It is the most famous dream in the history of science, and nobody other than Kekulé ever witnessed any part of it.
Two reveries, told once
Kekulé in fact described two visions that evening. The first belonged to his years in London, where he worked in the mid-1850s alongside the chemist Alexander Williamson. Riding home on a horse-drawn omnibus on a summer night, he drifted off and watched atoms pair up and link into chains. He credited that reverie with his structural theory, published in 1857 and 1858, which held that carbon forms four bonds and that carbon atoms can link to one another.
The second vision was the fireside one, set in his rooms in Ghent, where he was professor of chemistry from 1858.
Neither can be dated firmly. Accounts put the omnibus ride in 1854 or 1855. The historian Alan Rocke, writing in Angewandte Chemie for the benzene theory's 150th anniversary, places the fireside episode probably early in 1862, and adds the qualifier that this holds only if it happened as described. Other writers say 1861. All such dates are inferences from the 1890 speech. No one has produced a letter, notebook or diary entry from the time that mentions either episode.
One detail usually gets lost. Kekulé did not claim a dream in the full sense. In the passage Rocke quotes he says he "sank into half-sleep." Rocke's own word for it is daydream.
The case that he made it up
The benzene paper was read to the Chemical Society of Paris on January 27, 1865, not by Kekulé but by his friend Adolphe Wurtz. A longer German version followed in 1866. For a quarter of a century after that, Kekulé published nothing about snakes.
That silence is the core of the skeptical case, pressed hardest in the 1980s by John Wotiz, a chemistry professor at Southern Illinois University, and his collaborator Susanna Rudofsky. A folder of their papers from 1982 to 1984, now in the Science History Institute's archives, includes an article titled "Kekule's Dreams: Fact or Fiction?" Wotiz later edited a 1993 volume, The Kekulé Riddle. His argument was that the dream was invented, and that it served a purpose: a vision needs no acknowledgment of anyone else's work. The science writer Philip Ball, in an essay for Chemistry in Britain, quotes Wotiz's summary: "you don't have to give footnotes for a dream."
Ball adds an observation of his own. The two reveries, supposedly years apart, are told in strikingly similar language, with the atoms performing much the same dance in each. That is what one would expect from a storyteller shaping his material. It is also what one would expect from a man recalling two experiences of the same kind decades later.
A different objection came from the chemist Levi Tansjö, in a 1993 paper bluntly titled "The Fictitious Kekulé-dream." He argued that Kekulé's German word Schlange was a figure of speech for a winding chain of atoms, and that there was never solid ground for tying the chemist to a literal serpent.
Who drew the ring first
The footnotes Wotiz had in mind belonged to three men. The French chemist Auguste Laurent printed a hexagonal symbol in 1854. The Scottish chemist Archibald Scott Couper, who arrived at the theory of carbon chains independently in Wurtz's Paris laboratory and lost priority to Kekulé through a delay in publication, sketched two ring structures in 1858. And in 1861 the Austrian Josef Loschmidt published molecular diagrams in which benzene's core appears as a circle.
Rocke's reply is that none of these was the benzene ring in Kekulé's sense. Laurent's hexagon was not offered as a picture of a molecule's architecture. Couper's rings were not aromatic compounds. Loschmidt's circle was a placeholder for a nucleus whose structure he treated as unknown. What Kekulé did in 1865, on Rocke's account, was something none of them attempted: he built an empirical argument, drawn from the known behavior of benzene's derivatives, for one specific arrangement of six carbons.
Rocke also points out that by 1890 nobody, Couper and Loschmidt included, had challenged Kekulé's priority, and that the Berlin speech was a solemn performance by a frail man, not an after-dinner joke. He concedes the limit of the defense. A private mental event cannot be documented, so the story can be neither confirmed nor refuted.
There is one curious piece of earlier evidence. In 1886, four years before the speech, a spoof journal produced by German chemists, the Berichte der durstigen chemischen Gesellschaft, ran a cartoon of six monkeys gripping one another in a circle. Whether that shows the snake story was already circulating, or merely that chemists liked animal jokes, is a matter of interpretation.
A snake with a long history
The image was not new. The tail-eating serpent, the ouroboros, is one of the oldest emblems in alchemy. A well-known example sits on folio 279 of a Greek alchemical manuscript in Paris, copied in 1478 by Theodoros Pelecanos of Corfu from much older material.
It is tempting to see alchemy handing chemistry its central diagram through a sleeping man's mind. The evidence does not go that far. Kekulé spoke of a snake, not of an ouroboros or of alchemy, and Tansjö's reading would remove the serpent altogether. A chemist educated in the 1840s could well have known the symbol. That he was thinking of it is a later gloss.
What sleep can and cannot do
Could such a thing happen? Modern sleep research says something like it can, within limits.
In 2004 a team led by Ullrich Wagner reported in Nature on volunteers given a tedious number task that concealed a shortcut. More than twice as many discovered the hidden rule after a night's sleep as after an equal stretch awake.
In 2021 researchers at the Pitié-Salpêtrière Hospital in Paris tested the drowsy threshold itself. Of 103 participants working on a similar task, 83 percent of those who spent about a minute in the lightest stage of sleep later found the rule, against 31 percent of those who stayed awake and 14 percent of those who slid into deeper sleep.
Those results concern contrived puzzles and a rule already implicit in the task. The solutions also came later, after an average of 94 further trials, not as visions. They show that dozing can loosen a stuck problem, not that it delivers molecular structures. Kekulé seems to have understood the proportion. In the same speech, by Rocke's account, he was careful to say that science takes far more than dreaming. He had been turning aromatic compounds over for years before he turned his chair to the fire.