They Reversed Time in a Quantum Experiment — It Didn’t Go as Expected

Cordoba, Argentina, 1997. Patricia Levstein lowers a glass tube of orange powder into a magnet and starts a burst of radio pulses built to make time inside that powder run backwards. A moment ago, a single hydrogen nucleus in that powder was carrying a small magnetic signal. It has already leaked away into the ones packed around it and dropped into the noise. The way warmth leaves a cup. What she is turning around is not the powder. It is what the powder is made of. Single atomic nuclei the quantum scale some of that signal will not come home and nobody in any lab since has ever got it back the tube goes down into the middle of the magnet where the field is at its most even the sequence is already written so she starts it and lets the machine run it out the magnet hisses because it has to be kept cold to work at all the powder lies in the bottom of the tube a fine orange line along the glass she starts the outward half a green line stands up on the screen holds, sinks, and goes flat. She waits out the flat part, and nothing on that screen moves for the rest of it. Then the pulses turn around. Everything in that room is exactly where it was a minute ago. The only thing that has moved is a line on the screen, twice. That flat line was the signal leaking out of the nucleus it started in, and into every neighbor it could reach. The turn is the whole trick. Those nuclei tug at each other constantly, and radio at the right strength and rhythm makes every tug pull the other way instead. So the signal has to walk back the way it came, at the same speed through the same neighbors. She does not touch anything. There is nothing to touch. The sequence is written and the magnet is holding, and for the whole length of the return there's a flat green line and a hiss. Then the second half runs out, and the screen is still flat, and it stays flat long enough to be worth watching. Nothing about this should work, though nothing forbids it either. The equations that run those nuclei read exactly the same in either direction, and always have. It's that nobody has ever seen it happen. Cream has never come back out of coffee. A dropped glass has never climbed back onto the table. The only reason there's a magnet in that room is that a spoonful of powder is small enough and simple enough to be the one place you might The flat line lifts. It comes up out of the noise where a moment ago there was nothing. A hump climbing back where nothing had been. The signal walking home to the nucleus it left. It takes as long coming back as it took going out. The same few thousandths of a second run in reverse. She has it. And it's short. It comes back smaller than it went out. She measures how much is missing. What they publish is that the missing part comes from the one tug the pulses can't turn around. The part of the interaction the sequence leaves alone. And if you know where that leak comes from, you can go and fix it. It will take her group 25 years to find out that it's wrong. She isn't the first to be wrong about it. 27 years earlier, three men in an American laboratory had already done the first half of this. They put a solid magnet in and let it do what solids do. Whatever order you give the nuclei inside a solid, they hand it around among themselves until it's gone. It happens fast, and there's nothing subtle about it. You set it up, you look away, you look back, and the arrangement you made isn't there. So they hit it with a hard train of radio pulses. Strong enough and shaped right, a train like that makes a solid behave as though the sign the interaction between its nuclei had been turned over, and the spreading runs the other way. The arrangement climbed back out of nothing, out of a sample that had been sitting in that magnet the whole time doing nothing anybody could see. The paper that came out that year announced a broken rule. Everybody had been treating that rule as furniture. The spreading was final, the way the heat in your house is final. The words time reversal went into the title of the follow-up. The trick went into the standard toolbox of solid state magnetic resonance. If you can undo the thing that smears your signal, you can see chemistry you otherwise can't. And every single one of them saw what Lovstein saw. The signal never came all the way back. You got most of it back, and the longer you let it wander on the way out, the less you got. Everybody noticed, and everybody blamed the equipment. Pulses are never perfect, and magnets are never even. In the early 90s, Richard Ernst and his collaborators made the question sharper. Stop asking whether a whole sample's order comes back. Ask whether the signal comes home to the exact nucleus it started in. That's far harder to see, and it's what led a small group in Argentina put a number on what fails to return. By the end of the 90s, Levstein and Horacio Pastavsky had put it in print, two papers a few years apart, both saying the same thing. The time the echo survives is set by the interaction they were inverting. The leak wasn't theirs. The thing setting how fast the return fell apart was the same interaction they were turning around. Pastavsky has told an audience what came back at them, and he does it in his colleagues' voices. You are not doing this properly. Your power is probably too high. Your pulses are probably off. The crystal is not perfect. There are impurities in it. So they cleaned. They cut the pulses sharper. They went back over it the next year, and the year after that, going down the list of variables one at a time. And what came home was as short as it ever was. In his words, they came up against the same threshold. You can't disprove an accusation like that. Every time they cleared one variable, the accusation moved to the next one. So Pastavsky's group built the thing that could end it. 18 years after that first run, they published a machine of an idea. Instead of turning the tug between the nuclei around at full strength, scale it, run it at half, turn it down again, down again, until the reversal is barely a reversal at all. The knob is real. part of the interaction they turn around gets multiplied by a number they choose. Anywhere from nothing up to a half. At a half, it's as strong a reversal as the sequence can make. At nothing, there's no reversal at all, just the forward run twice over. They take the whole sequence through setting after setting on the same adamantane sample and read the decay off each run. Same powder, same magnet, same pulses, one number changed. They ran it on adamantane, a white powder that's nothing but a cage of carbon and hydrogen, standard sample in them. They built it as a trap for themselves. If the leak really is contamination, their pulses, their magnet, their sample, then it's sitting outside the tug they're turning around. Turn that tug down, and the contamination stays exactly where it is while everything else shrinks away around it. Whatever dirt was theirs would be the only thing still in the picture. So they run it at half, and read off how fast the signal loses its way home. Then at a quarter, and read it off again. Then lower. Then lower still, down to a setting where there's hardly any reversal left to speak of, and every setting gets its own number in the column. They ran it at setting after setting, and past a certain strength, turning that dial stopped making any difference at all how fast the powder lost its way home. Whatever was eating the return was the powder itself. Pastavsky had said out loud how good they had got. Their precision was on the order of 1%. Even that 1% of error, he says, wasn't enough to explain why their system wouldn't recover any better than it did. The nuclei were doing it to each other. The same tugging that lets a signal spread through a solid is the tugging that closes the road behind it. The thing being reversed and the thing preventing the reversal are one thing. What they had written down at the start was that the missing part came from the interactions they failed to turn around. That's what people write when they believe the fault is theirs and fixable. Unwrite it. They put a figure on it in 2022. Every solid has a clock inside it. Call it the gossip time, how long it takes one nucleus to have told its neighbors everything it knows. That time is a property of the material, like density. It comes out of how tightly those nuclei are packed, and it runs the same whether anybody is trying to reverse anything or not. It comes with the powder, and it cannot be tuned. What they measured is that a reversal survives a handful of those clocks, and then it's finished. That measurement puts it at about seven of them. It's a small number, it's set by the material, and getting better at the experiment doesn't raise it. In cobaltocene, one of the powders they worked with, seven of those clocks is about a thousandth a second 150 years ago a man in Vienna named Loschmidt told the physicist who had just explained why the world only runs forward that he couldn't be right because the laws underneath the explanation do not care which way you go reverse every velocity in a room full of gas said Loschmidt and the gas will unmix itself the physics allows it for more than a century nobody could touch that because nobody could reverse every velocity in anything Cordoba could that is what a tube of powder in a magnet is, a place where Loschmidt's demand can actually be met. Three physicists, Ehrman, Chepelyansky, and Chepelyansky, recently published a paper in the journal Entropy, arguing that the fight is still open. Nobody put anything in a magnet for it. It's a calculation on cold atoms held in a trap, and it says a quantum system, unlike anything classical, could in principle be run all the way back at full strength. The argument Loschmidt started is still running, but every calculation of that kind has to come back to the same place in the end. Somebody has to put something in a magnet, turn the pole around, and watch what walks home. That kind of work never makes the news on the day it happens. I went looking for it because that short hump kept bothering me, and whatever else I turn up down here goes out to the people subscribed to this channel. The people who did put it in a magnet published their result as a question, and then wrote the plainest sentence in this whole story. Once we reach the timescale of a process we cannot control, it appears as an ultimate, irreversible time. It doesn't stay in one powder either, four of them now, and they are not alike. Cyclooctatetraene carries one ring of five hydrogens per molecule, ferrocene carries two, adamantane is a cage of carbon, the echo fades in a different shape in each of them, and the paper puts that down to how much of the fading the spin dynamics is doing on its own rather than anything coming in from outside. Writing it up years later, the group called it an unbeatable limit and a sort of intrinsic decoherence rate and said it has a much deeper origin they meant deeper than their equipment let the signal wander further on the way out and the hump that comes back is lower further again lower again and each time it dies a little faster than the run before then it stops dying faster past that point the runs stop telling each other apart different settings same fall landing one on top of the other the more of the interaction they failed to turn around the faster it dies right up to a point past that point the rate stops depending on it at all that ceiling is the measurement it's the fastest the return can be made to die and nothing they ever did to the experiment moved it 25 years passed between that first short hump and the paper with the ceiling in it Podovsky spent most of them being told the experiment was dirty the paper says it wasn't