What Is the Double-Slit Experiment?

In 1989, a team at Hitachi filmed electrons arriving at a detector one at a time. Each landed as a single dot, in an apparently random place. After ten electrons the screen looked like noise. After seventy thousand, the dots had arranged themselves into stripes. That film captures the double-slit experiment, which Richard Feynman said holds the heart of quantum mechanics. It is also the source of one of the most repeated myths in science: that looking at a particle changes what it does. The real story is stranger and more precise, and it is about information, not observers.

The experiment in one picture

Fire something at a barrier with two narrow slits in it, and record where it lands on a screen behind.

With ordinary particles, such as tiny ball bearings, you get two bands on the screen, one behind each slit. With waves, such as ripples on water, you get something else. The waves spread from both slits and overlap. Where a crest from one slit meets a crest from the other, they add up. Where a crest meets a trough, they cancel. The result is a row of bright and dark stripes, called an interference pattern.

Now fire electrons, one at a time. Each arrives as a single, localised dot, exactly like a particle. Yet the dots pile up into interference stripes, exactly like a wave. Close one slit and the stripes vanish. Open both and they return, including dark bands where almost no electron lands, even though opening a second slit gave each electron more ways through. That last detail is the one no classical picture can explain.

The double-slit experiment, one electron at a time

Young’s sunbeam, and the card that started it

The experiment began with light. On 24 November 1803 Thomas Young gave the Royal Society’s Bakerian Lecture on optics. His setup was simpler than the textbook version. He let a thin beam of sunlight into a room through a hole in a window shutter, then split it with a slip of card about a thirtieth of an inch wide.

Coloured fringes appeared in the card’s shadow. Then Young made the observation that matters most. When he blocked the light passing one edge of the card, the fringes disappeared. They needed light arriving by both routes at once. That is the entire logic of the modern experiment, recorded in 1803.

The two-slit arrangement itself first appears in Young’s 1807 Course of Lectures on Natural Philosophy. The physicist and author Tony Rothman has argued that there is no clear evidence Young ever ran it in that form. Young’s wave theory met resistance at first. Once Augustin Fresnel extended the work in the following decade, physicists accepted that light is a wave, and for a century that seemed settled.

Then particles did it too

Quantum theory reopened the question by showing that light also arrives in lumps, called photons. In 1909 Geoffrey Ingram Taylor dimmed a light source so far that it delivered, on average, far less than one photon at a time. The fringes still formed on his photographic plate, given long enough.

Matter turned out to behave the same way. Electrons were shown to diffract in the 1920s, and in 1961 Claus Jönsson in Tübingen fired electrons through real, microscopic slits and recorded the stripes.

Feynman chose this experiment to open his 1960s lectures on quantum mechanics. He called it a phenomenon impossible to explain in any classical way, and added: “In reality, it contains the only mystery.” At the time, the single-electron version was widely treated as a thought experiment, too delicate ever to perform.

One electron at a time

It was performed, and the credit is tangled. In 1974 Pier Giorgio Merli, Gian Franco Missiroli and Giulio Pozzi in Bologna submitted a paper showing an interference pattern built from electrons arriving individually. It appeared in 1976 and attracted little attention.

In 1989 Akira Tonomura and colleagues at Hitachi published the version everyone remembers: a movie of single electrons accumulating on a screen, with frames at 10, 100, 3,000, 20,000 and 70,000 electrons. Their “double slit” was actually an electron biprism, a fine charged wire that splits the beam into two, which does the same job.

In 2002 readers of Physics World voted the single-electron double-slit experiment the most beautiful experiment in physics. It was the only entry in their top ten with no single name attached.

The rule behind the stripes

Quantum mechanics describes the electron with a wavefunction, and the wavefunction passes through both slits. Call the part coming through slit 1 ψ₁, and the part through slit 2 ψ₂. The probability of landing at a point is the square of their sum:

P₁₂ = |ψ₁ + ψ₂|² ≠ |ψ₁|² + |ψ₂|²

For ordinary particles, probabilities for two separate routes simply add. For quantum objects, the amplitudes add first and only then get squared. Expanding the square produces an extra cross term, and that term is the interference. Where ψ₁ and ψ₂ are in step, it adds probability. Where they are out of step, it subtracts, all the way to zero on the dark bands.

This is the Schrödinger equation at work, with the Born rule turning amplitudes into odds. Nothing in the rule refers to observers. It refers only to whether the two routes remain indistinguishable.

What “observing” really means

Put a detector at the slits to find out which one each electron uses, and the stripes vanish. The screen shows only a smooth spread, the plain sum of what each slit gives on its own, as if the electrons had been ordinary particles all along. This is the famous part, and the popular explanation, that a conscious observer changes reality by looking, is wrong.

What destroys the pattern is which-path information: any physical record, anywhere, that distinguishes the two routes. It can be a detector reading, a scattered photon, or a slight change in a nearby atom. Nobody has to read the record. It only has to exist.

The trade-off is exact. In 1996 Berthold-Georg Englert proved an inequality linking the visibility of the fringes, V, to how well the paths can be distinguished, D:

D² + V² ≤ 1

Perfect path information means no fringes. No path information allows perfect fringes. Partial information gives faded fringes, in between. A 2008 experiment with electrons in a tiny ring made the point directly: interference was suppressed only when path information could be acquired, even if only in principle.

Which-path information fades the stripes

Information, not a kick

The older textbook explanation goes back to Niels Bohr and Werner Heisenberg. Measuring which slit the electron used, the argument runs, gives it an unavoidable kick. The kick scrambles its momentum and smears out the fringes. Disturbance does the damage.

Experiments have since separated the two ideas. In 1998 Stephan Dürr, Tilman Nonn and Gerhard Rempe at Konstanz marked which path atoms took by changing their internal state, with a momentum kick far too small to wash out the fringes. The fringes vanished anyway. Later analysis summed up the lesson: the experiment decided against momentum transfer and in favour of entanglement between each atom’s path and its marker.

Englert’s inequality points the same way. His derivation never uses the uncertainty principle at all. A kick can destroy fringes, but it is not the fundamental reason they disappear. The fundamental reason is that the two routes have become distinguishable.

The quantum eraser, and the myth of changing the past

If information destroys the fringes, can erasing the information bring them back? Marlan Scully and Kai Drühl proposed exactly that in 1982, and it works. Mark the paths, then scramble the marker so it no longer reveals the route, and interference reappears.

The famous version is the delayed-choice quantum eraser, published in 2000 by Yoon-Ho Kim, Marlan Scully and colleagues. It uses pairs of entangled photons. One photon of each pair hits a detector that records where it lands. Its twin travels further and meets a random choice, at a beam splitter, that either preserves or erases its path information. That choice happens after the first photon has already landed.

Popular accounts claim this lets a later choice reach back and change the past. It does not. That detector’s record, taken on its own, never shows fringes. Fringes appear only when the landings are sorted afterwards into groups, using the twins’ results, which must be brought over by ordinary means. The two “erased” groups show fringes and anti-fringes that exactly cancel when added back together. The choice selects which subset you look at. It sends nothing backwards in time.

Einstein’s recoiling slit, tested with single atoms

Albert Einstein tried to beat the rule at the 1927 Solvay conference. Suspend one slit on a delicate spring, he suggested. A particle passing through would nudge it, revealing its path, while the fringes still formed. Bohr replied that measuring the nudge precisely enough would blur the slit’s own position, and the fringes would wash out.

In 2025 Wolfgang Ketterle’s group at MIT built the cleanest version yet, published in Physical Review Letters. They held more than 10,000 atoms, cooled to millionths of a degree, in a lattice of laser light. Each atom served as a slit, which Ketterle described as “the smallest slits you could possibly build.” Light was kept so weak that each atom scattered at most one photon. By tuning how tightly the atoms were held, the team controlled how much path information the atoms picked up, and the fringes faded exactly as quantum mechanics predicts. Switching off the trap, removing Einstein’s spring, changed nothing.

Headlines announced that MIT had proved Einstein wrong and settled a century-old debate. That overstates it. Bohr answered Einstein in 1927, the quantum prediction was never in serious doubt, and experiments in the 1990s had already tested the principle. What MIT added was the most idealised implementation ever built.

How big can the “particle” be?

The same experiment works with objects far bigger than electrons. In 1999 Markus Arndt and Anton Zeilinger’s group in Vienna sent football-shaped molecules of 60 carbon atoms through a grating and saw interference. By 2019 the group had done it with molecules of about 2,000 atoms. In January 2026 they reported clusters of more than 7,000 sodium atoms.

The obstacle to going bigger is the one this article has been circling. A large, warm object constantly exchanges light and heat with its surroundings, and every exchange risks leaking which-path information. Once that information exists anywhere, the fringes go. That leakage is decoherence, and it is why the experiment gets harder as objects grow. It is also why nobody has ever seen Schrödinger’s cat go through two slits.

Established, contested, unproven

Established. Single photons, electrons, neutrons, atoms and molecules all build interference patterns one quantum at a time. Which-path information and fringe visibility trade off exactly, as Englert’s inequality states. No conscious observer is required: an unread record is enough. Quantum erasers do not send information into the past.

Contested. What the electron is doing between the source and the screen. In standard quantum mechanics the wavefunction passes through both slits and the question of a single path has no answer. In pilot-wave theory the electron takes one definite path, steered by a wave that passes through both. Many-worlds offers another account again. All reproduce the same stripes. Even the phrase “wave-particle duality” is disputed, since many physicists prefer to say quantum objects are neither.

Unproven. Whether interference continues for arbitrarily large objects. Some theories predict it must fail beyond a certain mass. Every larger experiment narrows where that limit could be, and none has found it.