The Mpemba Effect: Does Hot Water Freeze Faster Than Cold?

On its own, “does hot water freeze faster than cold water?” isn’t answerable. Too much is left unstated. Before you can test it you have to fix the starting temperatures, what the water is made of, the shape and material of the container, what surrounds it as it cools, and the slipperiest variable of the lot, what you are willing to call “frozen.”

Take that last one. “Frozen” might mean the first ice crystal forming, or the surface reaching 0°C, or a thermocouple in the middle reading 0°C, or the whole volume going solid. Four separate moments, and in the same beaker they do not always arrive in the same order. Score the race by one of them and the hot sample wins; score it by another and the cold one does. People have been talking past each other over that single ambiguity for more than two thousand years.

Stated carefully, the real question runs: under what reproducible conditions does water that starts hotter reach a fixed freezing endpoint before an otherwise identical sample that starts colder? Put that way, there is no clean yes or no. Sometimes it happens, sometimes it doesn’t, and even where it does, the physics behind it is only partly worked out.

Aristotle, in Meteorologica (Book I, Chapter XII, c.350 BCE), put it plainly: “The fact that the water has previously been warmed contributes to its freezing quickly: for so it cools sooner. Hence many people, when they want to cool water quickly, begin by putting it in the sun.” The explanation he reached for, antiperistasis (a quality supposedly intensifying when hemmed in by its opposite), is wrong by any modern reading. What he reported seeing has held up better than his theory of it.

The note keeps recurring. Francis Bacon, in Novum Organum (1620): “Slightly tepid water freezes more easily than that which is utterly cold.” René Descartes, in the Discourse on Method (1637), claimed water left long on a fire froze faster, and pinned it on the evaporation of those particles “least able to resist bending.”

The modern thread begins in 1963. Erasto Mpemba, a Form 3 student at Magamba Secondary School in Tanzania, was making ice cream with his class. Short on freezer space and out of time, he pushed his mixture in while it was still hot from boiling, rather than waiting for it to cool the way the others did. About ninety minutes later his batch had frozen solid while theirs was still liquid. His physics teacher told him that was impossible, and “Mpemba’s physics” became the joke around school.

The question stayed with him. At Mkwawa High School, when the physicist Denis Osborne came to give a lecture, Mpemba stood up and put it to him directly: “If you take two similar containers with equal volumes of water, one at 35°C and the other at 100°C, and put them in a refrigerator, the one that started at 100°C freezes first. Why?” Osborne could have brushed it aside. He admitted instead that he did not know, and promised to check. Back in Dar es Salaam, his technician ran the test, and the result came back the way Mpemba had described.

Their joint paper ran in Physics Education in 1969, under the title “Cool?” With 70 ml of water in 100 ml beakers, stood on polystyrene inside a domestic freezer, the samples that reached the onset of freezing fastest were the ones starting near 90–100°C. Evaporation, they worked out, could account for no more than about 30% of the cooling, which left something else to do the rest.

The interest never quite died down. In 2012 the Royal Society of Chemistry put up £1,000 for the best explanation and drew more than 22,000 entries, whittled to eleven finalists by expert judges and a public vote. The prize went to Nikola Bregović of the University of Zagreb, announced in January 2013, for an account built on supercooling and convection. Awarding it settled who had explained the effect best that year, which is not the same as settling what causes it.

Newton’s Law of Cooling says a body sheds heat at a rate set by the gap between it and its surroundings: dQ/dt ∝ ΔT. So the hotter sample loses heat faster at the start, while that gap is widest. The catch is that it also has more heat to get rid of on the way down to 0°C. Run the figures as though nothing else about the sample changes, and the colder one reaches freezing first every time, by a comfortable margin. It is plain bookkeeping.

By that logic, hot water beating cold water to ice is simply impossible. Burridge and Linden (2016) point to where the logic leaks: it assumes heating only pours in energy and leaves everything else alone. Heating is never that clean. Boiling a sample drives off some of its mass, pushes out dissolved gas, builds stronger internal currents, moves the spots where ice first takes hold, and even changes how the vessel meets the freezer shelf. The warm sample going into the cold is not the cold sample with extra joules added. It has become a system of its own, and once two different systems are racing, the one the equation favours can still come second.

None of these rules out the others. In one experiment two or three can line up and reinforce each other, or pull against each other and partly cancel. That is what makes the results so touchy about conditions, and why no single mechanism has ever been signed off as the full answer.

Start with the evidence that holds. The 1969 paper was a set of repeated, documented measurements in a refereed journal, well past anecdote. Brownridge (2011, American Journal of Physics) pushed it further: cycling a pair of sealed vials through 28 freeze-thaw runs, the hotter one froze first every time. The catch is that he had picked those vials because their nucleation temperatures differed by about 5.5°C, so the two samples were not identical to begin with, and the supercooling gap is doing the work. Burridge and Hallstadius (2020) got a repeatable effect by roughening the inside of the hot sample’s container while leaving the cold one smooth, which is about as close as anyone has come to naming a mechanism in a setup built to show it. Both results need two things to differ between the samples, container and temperature, not temperature alone.

A 2023 paper in InfoMat reported watching the effect happen directly in water under set conditions. The most recent work, Janni et al. in RSC Applied Interfaces (2026), ran 176 freezing measurements and caught the effect in somewhere between 46 and 58 of them depending on how freezing was scored, tracing it to fluctuations in convective air currents and differences in supercooling. In those rigs, it happens.

The same behaviour shows up well outside the kitchen, in tetrahydrofuran clathrate hydrates, in granular fluids, in magnetic minerals, even in quantum systems. That spread hints that water isn’t really the point. The deeper pattern, a system relaxing faster from a hotter start under the right conditions, looks like a property of non-equilibrium physics in general, with water as just one place it surfaces.

Now the other side. Burridge and Linden (Cambridge, Scientific Reports, 2016) built tightly controlled experiments, fixed “freezing” as cooling to 0°C, and found no reliable effect at all. Several earlier positive results, they showed, were artefacts of how the measurement had been arranged. Misplace the thermometer vertically by as little as a centimetre and you can manufacture an apparent Mpemba Effect out of nothing, because the water column isn’t one temperature top to bottom.

Their wider point is the harder one to dodge. Pin freezing to 0°C, control the rest, and the textbook prediction holds up fine. The positive findings cluster in exactly the cases where temperature isn’t the only thing differing between the two samples. Something real may be happening in those rigs; it just isn’t something you can lay on temperature by itself.

The defensible answer is a narrow one. The Mpemba Effect is real, but conditional. It is neither the dependable rule the “hot freezes faster” headlines suggest nor the impossibility its thermodynamic critics keep declaring it to be. What it amounts to depends almost entirely on the setup, and the setups where it appears share a profile:

It favours rough, kitchen-like conditions. A starting gap of roughly 30 to 50°C between the samples. Open containers, so evaporation can play a part. Ordinary tap water carrying dissolved minerals, not deionized water. A home freezer running a little below freezing, somewhere around −5 to −10°C. And freezing judged by the first ice to form rather than by the whole sample turning solid.

Clean the experiment up and it fades. Place the containers identically, cap them so nothing evaporates, use deionized and degassed water, define freezing as the centre hitting 0°C, and keep nucleation under control, and the advantage mostly goes away.

No one mechanism covers every case. Across most rigs where the effect does appear, the bulk of the evidence puts convection and supercooling differences in the lead, with evaporation adding to it whenever the container is open. The hydrogen-bond idea is the interesting outlier, suggestive but still unproven at any scale you could measure in a beaker.

Sixty years of modern work, on top of centuries of scattered observation, still haven’t produced one accepted theory that covers every Mpemba result. The candidate mechanisms are understood well enough. What is missing is a model that can say in advance which conditions will produce the effect, and how strongly. Whether hydrogen-bond restructuring plays a part is unsettled, and so is the exact link between a container’s shape, where nucleation tends to begin, and how often the effect shows at all.

A phenomenon this dependent on its own setup was never likely to reduce to a single tidy law, and to its credit the literature mostly admits as much. Sixty years in, the unanswered pieces are simply what a difficult problem costs to keep working on.

The episode is a small study in how a field absorbs a result that supposedly can’t happen. The lazy response in 1963 was the one Mpemba’s teacher gave: thermodynamics forbids it, end of conversation. The slower response is to state the claim exactly, strip out the confounding variables, rerun it carefully, and keep whatever still stands. Osborne did the slow version. What stood up did so in some tests and fell over in others, and that is more or less where the matter rests today.

It also shows what reproducibility even means for a phenomenon this touchy. When two labs use different container materials, different water, different freezer temperatures, and different definitions of freezing, they are not running the same experiment, whatever the shared title says. Each can land a contradictory answer that is correct on its own terms. Calling that a reproducibility crisis inverts it. With a process this sensitive and several mechanisms firing together, careful labs disagreeing is the expected outcome, not the alarming one.

The definitional problem belongs at the centre of the story, not in a footnote. Popular write-ups almost always pose the effect as a yes-or-no. The truthful answer reads more like: yes, under this definition of freezing and these conditions; no, under that definition and those. A piece that never says what it means by “freezing” hasn’t really engaged the question.

This reaches past water. The same anomalous relaxation has been recorded in clathrate hydrates, in granular fluids, in quantum systems. The underlying physics is a good deal bigger than a story about ice cream.

The practical payoff is small. Ice cream makers really do start from warmed mixtures, and temperature choices matter in industrial freezing and in resurfacing a rink. None of that turns “hot water sometimes wins in particular rigs” into advice you can trust at your own kitchen sink.

The hydrogen-bond idea is worth watching, confirmed or not. If heating really does reorganize water’s hydrogen-bond network in a way that lingers through cooling and steers nucleation, that would be a new mechanism with reach well past ice formation. The question is still open in molecular physics.