The Feynman Slit
Fire particles at a pair of slits one at a time and the landings pile up into an interference pattern. It is stable, it is the same every run, and if you wipe it and keep firing it comes back. Every symptom of a basin, and not one of them earned.
This is the most seductive false positive in the column, because everything a basin is supposed to look like is here at once. A structure appears out of a formless scatter. It is sharp, reproducible, and indifferent to how you started. Damage it and it re-forms. The claim this page refuses is not that the physics is shallow, it is the deepest experiment there is. The claim it refuses is that the picture on the screen is a state being pulled home.
Every figure and quotation is sourced in the list at the bottom, with the date it was read. The panel partway down is a sampler drawing from the textbook distribution, labelled as such where it sits.
What actually happens
Send electrons at a barrier with two slits, slowly enough that only one is in flight at a time, and each arrives as a single dot. Feynman puts the discreteness first, before any talk of waves: "Electrons always arrive in identical lumps." Nothing smeared ever lands. What lands is one particle, at one place, once.
The wave is in the bookkeeping, not in the arrival. "The probability of an event in an ideal experiment is given by the square of the absolute value of a complex number φ which is called the probability amplitude." Two slits give two amplitudes, the amplitudes add, and the square of the sum has fringes in it. That squared amplitude is a distribution over where the next lump will land, and it is fixed by the apparatus: the slit spacing, the wavelength, the geometry. It is settled before the first electron is fired and it does not change while you fire.
This has been done at genuinely one-at-a-time rates. Bach and colleagues ran the electron source so dim that "the electron detection rate in the pattern was about 1 Hz" and built the figure up "by recording single electron detection events diffracting through a double-slit". Tonomura's group had shown the same buildup from individual dots in 1989.
the apparatus fixes a distribution → an electron is fired → it lands somewhere, drawn from that distribution → the dot is recorded → the apparatus fixes the same distribution
Read that loop again and notice what is missing. The recorded dots appear in it exactly once, at the end, and nothing downstream of them comes back around.
Why it looks like a basin
It self-assemblesearly on the plate is a meaningless scatter. Keep going and bands emerge that nobody drew. Order appearing from disorder with no hand shaping it is the signature this whole laboratory is built to look for.
It is the same every timerun it tomorrow, on a different machine, with a different operator, and you get the same fringes in the same places. Reproducibility of that quality usually means something is holding the pattern.
It does not care how you startthere is no initial condition to get right. Any starting scatter of early dots ends in the same picture, which is exactly what a wide basin of attraction looks like from the outside.
It recoversdeface the record, keep firing, and the fringes come back. This is the one that does the damage, because recovery after a disturbance is the test most people stop at.
Why it is refused
Nothing reads itthe landing position of the ten-thousandth electron is drawn from the same distribution as the first. No term anywhere in the physics takes the accumulated picture as an input. The electron in flight has no access to the plate's history and could not act on it if it did. That is the whole refusal: there is no loop closing on the structure's own prior state, so there is no recursion to lock, and the column's discriminator is exactly that.
It is not repaired, it is outvotedscrub a thousand dots off the plate and fire a thousand more. The scrubbed dots do not come back. They are simply diluted by new samples until the damage stops mattering, and the rate at which it stops mattering is the ordinary one over root N of a running average. A basin restores a state. An average drowns a disturbance in fresh evidence. Those look alike on a graph and are not the same event.
The recovery is bought, not producedand this is the part that settles it. Damage the record and stop firing: the damage stays, forever. Nothing in the apparatus notices or corrects it. The pattern only returns while an outside agent keeps paying for fresh draws, which makes the restoring work the experimenter's, not the system's.
The test that separates them
The confusion is that two very different interventions both get called a disturbance.
| What you disturb | What happens | What it shows |
|---|---|---|
| The record: wipe dots off the plate | The pattern re-forms, but only while the source keeps firing | Nothing. You damaged a tally and then paid for a new one |
| The system: close a slit | The distribution becomes the single-slit one at once, and stays there | There is no restoring force. The distribution is whatever the apparatus currently specifies |
The second row is not hypothetical. Bach's apparatus put "a movable mask in front of a double-slit to control the transmission through the individual slits", so both arrangements were observed in the same instrument. Closing a slit does not provoke a return to the two-slit fringes. There is nothing there that wants them.
That is the general shape of the test, and it is worth carrying to the other cards in this column: a basin returns when you perturb the system. This returns only when you perturb the bookkeeping and keep buying samples.
The same test, run here
Electrons arrive one at a time. Wipe the marked band and watch what the picture does while the source is firing, then stop the source and wipe it again. Then close a slit, which is the other kind of intervention entirely.
Deviation of the tally from the distribution the apparatus specifies, against electrons fired, on log axes. The gold line is the deviation expected from independent draws at the number of dots actually on the plate, which is the level damage decays toward rather than a prediction of the deviation right now: straight after a wipe or a moved slit the plate is not a clean sample of anything. It steps up when you wipe, because throwing dots away leaves fewer of them to average over. Blue ticks mark a damaged record, red ticks a moved slit.
sinc²(Au)·cos²(Du),
normalised over the plate; closing one slit removes the interference factor
cos²(Du), and the single-slit envelope that remains is renormalised over the
plate in turn, so the panel is always sampling where an electron lands given that one arrived,
never how many arrive. No quantum state is
evolved, because the page's claim does not need one. The independence of successive arrivals
is not demonstrated here, it is taken from the cited experiments, and the falsifier above
says exactly what measurement would overturn it. What the panel shows is what accumulation
does given that independence, which is the only thing in dispute.
What the trace is saying
Left undisturbed, the measured deviation walks down the gold line. That line is sampling noise and nothing else: a tally of N independent draws sits about one over root N away from the distribution it came from, and that is all the stability in the picture.
Wipe the band while firing and the deviation jumps, then falls back to the line on a steeper path, roughly one over N. That steeper leg is the whole argument in one measurement. A missing block of dots is a fixed deficit, and a fixed deficit divided by a growing total shrinks like one over N. It is arithmetic done by the incoming samples, not relaxation by anything in the apparatus, and it stops the moment you stop paying for draws.
Stop the source first and the deviation does not move at all, for as long as you care to wait.
Close a slit and the deviation jumps for the opposite reason: the plate has not changed, the distribution has. The old tally is now wrong about a new apparatus and gets outvoted on the same one over N schedule. At no point does anything pull the two-slit fringes back.
What would show this reading wrong
If the landing position of one electron were measurably influenced by where earlier electrons landed, with the apparatus held fixed, there would be a loop closing on the pattern's own prior state and the refusal would fail. This is a real measurement, not a thought experiment: it is a correlation between successive detections at fixed settings.
If a damaged pattern repaired itself with the source switched off, the restoring work would be the system's and the refusal would fail.
If closing a slit produced a transient drift back toward the two-slit fringes before settling, rather than an immediate switch to the new distribution, there would be something holding the old pattern.
Honest limits
Nothing here disputes the physics or takes a side between interpretations of quantum mechanics. The interference is real, the amplitudes add, and Feynman's judgment that the experiment holds "the only mystery" is not being argued with. The refusal is narrow: it is about the picture accumulating on the screen, which under any interpretation is a tally of detections.
The quantum state itself does have dynamics. It evolves smoothly and reversibly, and that evolution is the most successful law in physics. The pattern on the plate is not that state. It is a record of outcomes drawn from it.
Strict independence between arrivals is an idealisation. At high beam currents there are space-charge and source effects that correlate electrons. That is one reason the one-at-a-time experiments run the source down to about one detection per second, and it is also why the falsifier above is worth taking seriously rather than treating as settled.
The figures and quotations are a reading of published work. Nothing on this page was measured here. The panel is a sampler running the textbook distribution, not an experiment and not evidence for the independence it assumes, and it says so on itself.
Sources
Feynman, R. P., Leighton, R. B., & Sands, M. The Feynman Lectures on Physics, Volume III, Chapter 1, "Quantum Behavior". Source of the three quotations: the only mystery, electrons arriving in identical lumps, and the probability amplitude. feynmanlectures.caltech.edu/III_01.html
Bach, R., Pope, D., Liou, S.-H., & Batelaan, H. (2013). Controlled double-slit electron diffraction. New Journal of Physics, 15, 033018. Source of the movable mask, the roughly 1 Hz detection rate, and the buildup from single detection events. doi.org/10.1088/1367-2630/15/3/033018
Tonomura, A., Endo, J., Matsuda, T., Kawasaki, T., & Ezawa, H. (1989). Demonstration of single-electron buildup of an interference pattern. American Journal of Physics, 57(2), 117–120. Cited for the single-electron buildup, not quoted; the publisher's page returned 403 and no full text was reachable. doi.org/10.1119/1.16104
Related on this site: False Basins, the column this belongs to; the Clock-Driven Pattern, the same refusal from the opposite direction, where the recovery is perfect because nothing reads the state; and Finding a Basin for the tests this page applies.