Does Hot Water Really Freeze Faster Than Cold Water? The 60-Year Truth About the Mpemba Effect
Every winter, somewhere in the world, someone fills two ice-cube trays — one with hot tap water, one with cold — slides them into the freezer, and waits. The kitchen folklore says the hot tray wins. The claim has a name, a history stretching back more than two thousand years, and a reputation as one of physics’ most delicious paradoxes: the Mpemba effect, the assertion that hot water can freeze faster than cold water.
It sounds like it should be impossible. It sounds like the kind of thing a science teacher would confiscate. And indeed, teachers did try — the effect is named after a schoolboy whose own physics teacher dismissed his observation as nonsense. Yet the claim has refused to die. It has survived peer-reviewed papers, a thousand-pound prize for its explanation, and decades of increasingly careful experiments. So what is actually going on? Does hot water really freeze faster than cold, or is this one of science’s most persistent half-truths?
The honest answer is more interesting than either the viral version or the flat denial. Hot water has genuinely been observed to freeze before cold water — repeatedly, measurably, under specific conditions. But the way the claim usually travels, as a general law of nature, is wrong. That is what makes it a perfect fact-check: a real phenomenon, wrapped in a misleading story.
The claim, in its viral form
As it circulates online and in classrooms, the claim is blunt: hot water freezes faster than cold water, full stop. The story usually comes with a pedigree. It is named for Erasto Mpemba, a Tanzanian schoolboy who noticed it in 1963, and it carries the romance of the outsider who embarrassed the experts. Articles and videos present it as a settled oddity of physics — one of those delightful cases where common sense is wrong and the universe is stranger than you think.
The implication, whether stated or not, is that this is a general rule. Fill two identical containers, one hot and one cold, put them in the same freezer, and the hot one will reliably turn to ice first. Some versions go further, suggesting you should fill your ice trays with hot water to get ice sooner, or that the effect reveals something deep and mysterious about the nature of water itself.
To see why physicists were skeptical, consider the logic that the claim appears to violate. Newton’s law of cooling holds, roughly, that the rate at which an object loses heat is proportional to the temperature difference between the object and its surroundings. The hot sample starts further from the freezing point. It must pass through every temperature the cold sample passes through — including the cold sample’s starting temperature — before it can freeze. So how could it possibly arrive first? It is like claiming that a runner who starts a race ten meters behind the starting line can cross the finish line first without running any faster. The claim seems to breach basic thermodynamics, which is precisely why it has fascinated researchers for over half a century.
And yet observations resembling the effect are not new. They have been noted, on and off, for millennia.
The ancients got there first
Long before anyone had a freezer, philosophers were already arguing about warm water and ice. Aristotle, writing in the 4th century BCE, recorded the observation that water which had been warmed seemed to freeze more quickly. In one common translation of his Meteorologica, he wrote: “The fact that the water has previously been warmed contributes to its freezing quickly: for so it cools sooner.” People of his time, he noted, would put water out in the sun when they wanted it to cool quickly — a practice that only makes sense if you believe warming first somehow accelerates the later cooling.
The idea kept resurfacing. In the 17th century, Francis Bacon discussed similar observations, and René Descartes wrote in his Discourse on the Method that “one can see by experience that water that has been kept on a fire for a long time freezes faster than other,” offering his own explanation involving the particles driven off by heating. For centuries, the observation lived in that strange territory between folk wisdom and natural philosophy: reported by serious thinkers, explained by nobody to anyone’s satisfaction, and never quite subjected to the kind of controlled test that could settle it.
That changed because of a teenager in Tanzania who refused to let the matter drop.
A schoolboy, an ice-cream maker, and a question no teacher would answer
The year was 1963. Erasto Mpemba was a Form 3 student at Magamba Secondary School in what was then Tanganyika. In cookery class, the students made ice cream by boiling milk, mixing in sugar, letting it cool, and then freezing it. Freezer space was limited, so it was first come, first served. Rushing to claim a spot, Mpemba put his mixture into the freezer while it was still hot, skipping the cooling step his classmates followed. To his surprise, his hot mixture froze into ice cream before the cooled mixtures did.
He asked his teacher to explain it. According to the account Mpemba later co-wrote, the teacher dismissed the result — joking that it was “Mpemba’s physics,” not real physics. The phrase stuck; his classmates used it to mock his questions. But the puzzle stuck with Mpemba even harder.
The breakthrough came years later, almost by accident. Mpemba had moved on to Mkwawa Secondary School in Iringa when the headmaster invited Dr. Denis Osborne, a physicist from the University College in Dar es Salaam, to give a guest lecture. During the question session, Mpemba stood up — to the ridicule of his classmates — and asked: “If you take two similar containers with equal volumes of water, one at 35°C and the other at 100°C, and put them into a refrigerator, the one that started at 100°C freezes first. Why?”
Osborne assumed the boy was mistaken. But instead of brushing him off, he went back to his laboratory in Dar es Salaam and tried the experiment himself. To his amusement, the strange result held up. In 1969, the physicist and the former schoolboy published their findings together in the journal Physics Education, in a paper titled, fittingly, “Cool?” By then Mpemba was studying at the College of African Wildlife Management, and the phenomenon carried his name.
Their paper reported careful work. Mpemba and Osborne placed 70-milliliter samples of water in 100-milliliter beakers in the ice box of a domestic refrigerator, set on a sheet of polystyrene foam, and timed how long each took to begin freezing. They found the wait was longest for water starting at around 25°C and much shorter for water starting near 90°C. They checked the obvious objections: they ruled out the loss of liquid volume through evaporation as a significant factor, and they ruled out the effect of dissolved air. Most of the heat loss, they found, came from the liquid’s surface. Osborne later described the setup plainly: hot water in Pyrex beakers on polystyrene foam in a domestic freezer froze before cooler samples, and he and Mpemba attributed this to convection keeping the top of the hot sample warm and losing heat rapidly.
The schoolboy had been right — at least under the conditions of that experiment. But what exactly had he been right about? That question would consume the next six decades.
Why it shouldn’t work: the skeptical case
The most forceful modern challenge to the Mpemba effect came from Cambridge. In 2016, physicists Henry Burridge and Paul Linden published a long paper in Scientific Reports with a title that left little to the imagination: “Questioning the Mpemba Effect: Hot Water Does Not Cool More Quickly Than Cold.” They surveyed the confused history of the subject, ran their own experiments under carefully controlled conditions, and concluded, in their abstract’s own words, that there was “no evidence to support meaningful observations of the Mpemba effect.”
Their methodological point was sharp. Much of the confusion, they argued, comes from supercooling — the fact that water can remain liquid below 0°C before ice crystals finally nucleate. When experimenters time “freezing,” they are often timing a messy combination of cooling and the unpredictable moment of nucleation. Burridge and Linden sidestepped the mess by measuring something cleaner: how long it took water samples to cool down to 0°C, excluding supercooling from the picture entirely. Measured that way, the effect vanished. Hot water did not reach 0°C faster than cold water. The effect, they suggested, was an artifact of how freezing had been defined and timed.
They were not alone. MIT’s School of Engineering gives the question a crisp treatment. “It’s an age-old question with a simple answer: no,” says mechanical engineer Prakash Govindan. His reasoning is the runner analogy made formal: “No matter what the initial temperature of water is, it must be brought to the freezing point before it will change state and become ice. It will actually take more time and/or energy to freeze hot water because it must be brought down further in temperature until it reaches the freezing point, about 0°C.”
Govindan does concede one wrinkle. Put hot and cold containers in the same freezer, and the freezer works harder on the hot one — the initial rate of heat transfer is faster for the hotter sample. But both containers are cooling simultaneously, and the cold one keeps its head start all the way down. Under standard conditions, the logic of the skeptics is airtight: nothing about being hotter helps you finish a cooling race sooner, because you still have to travel the extra distance.
Chemistry commentators have echoed the point. A detailed review of the freezing and cooling literature concluded that the Mpemba effect “is not supported as a true physical phenomenon” under standard conditions, arguing that studies measuring the time for water to reach freezing under atmospheric pressure find no evidence for it. In 2023, Skeptical Inquirer ran a long retrospective titled “The Rise and Fall of the Mpemba Effect,” arguing that poor reporting, vague methods, and a lack of emphasis on reproducibility had let the myth live far longer than it deserved.
If the story ended there, the verdict would be simple: False. But the story does not end there — because other careful experimenters kept getting the opposite result.
Why it sometimes does: the experiments that found it real
The strongest affirmative evidence comes from James Brownridge, a physicist at the State University of New York at Binghamton, who spent years chasing the effect with unusual patience. Brownridge reported repeating his experiment under tightly described conditions and getting the same result every time: in 28 attempts, hot water freezing before cold water was observed 28 times. His apparatus used a vacuum pump to cool samples starting at different temperatures, and he found something genuinely strange — the warmer water not only cooled faster, it kept cooling faster even after its temperature had dropped to match the colder sample’s. With optimal starting temperatures, the two cooling curves crossed at 2.7°C: below that point, the initially hotter water was actually colder than the initially cooler water. That is not a measurement artifact. That is a real, reproducible crossover.
Brownridge’s work also clarified the single most important condition for the effect. As Nikola Bregović summarized in his prize-winning essay on the subject, quoting Brownridge: “Hot water will freeze before cooler water only when the cooler water supercools, and then, only if the nucleation temperature of the cooler water is several degrees lower than that of the hot water. Heating water may lower, raise or not change the spontaneous freezing temperature.” In other words, the effect is real, but it is fragile. It depends on the accident-prone physics of supercooling — whether and when each sample happens to nucleate ice — and heating the water changes that nucleation behavior in ways that are not fully predictable.
A different line of attack reached a compatible conclusion. Physicists Andrei Klimov and Alexei Finkelstein ran water-freezing experiments under carefully controlled conditions and found that the Mpemba effect appeared only when the freezer temperature sat very close to the temperature at which ice nucleation begins. Under those conditions, the range of freezing times for both hot and cold samples was enormous — so enormous that the spread exceeded the hot sample’s cooling delay, and the hot water sometimes froze first. Their theoretical analysis traced the paradox to the stochastic nature of ice nucleation, the randomness inherent in a first-order phase transition. The “effect,” in this view, is not hot water breaking the laws of thermodynamics. It is statistics: when freezing times are wildly variable, the sample that started hotter can win the lottery.
Even students have reproduced it. A 2016 paper in the Institute of Physics journal Physics Education described a Cambridge Pre-U student’s investigation using small aluminum vessels in a freezer held between −19.1°C and −18.8°C — and the student found the effect, then repeated the experiments eighteen months later with the same success. The author was careful to note that there was still no consensus on a definitive explanation. But the observation itself, in that apparatus, held up.
The £1,000 question
By 2012, the debate had been running, unresolved, for more than four decades. The Royal Society of Chemistry decided to force the issue: it offered a £1,000 prize for the best and most creative explanation of the Mpemba effect. The competition drew 22,000 entries from around the world, whittled down by expert judges and a public vote to just 11 finalists. The winner was announced on January 10, 2013 — and Erasto Mpemba himself was there for the announcement, appearing by live video link as the society declared the victor in London and Zagreb.
The winner was Nikola Bregović, a research assistant in physical chemistry at the University of Zagreb in Croatia, who had been tipped off to the contest by a friend’s email and began experimenting in his mentor’s lab almost immediately. His winning essay confined itself to elementary thermodynamics and pointed to four factors of possible significance: evaporation, dissolved gases, convection, and supercooling. His conclusion, leaning on Brownridge’s work, was that supercooling was ultimately the significant factor — the hot water wins only when the cold water supercools and nucleates at a sufficiently lower temperature.
Note what the competition did not produce: a consensus. Bregović won for the best explanation, not the final one. The RSC’s own education resources still present the effect as an open invitation — try the experiment yourself, read the original paper, join the argument. More than a decade after the prize was awarded, researchers were still publishing new analyses of when and why the effect appears.
The candidate explanations
Over the years, scientists have proposed a whole menu of mechanisms, and most of them are probably contributing in different setups:
Evaporation. Hot water evaporates faster, so there is simply less of it left to freeze. Mpemba and Osborne tested this and ruled it out as a significant factor in their setup, but in open containers over long cooling times, mass loss is real.
Convection. Hot water churns. The resulting currents keep the surface hot — Osborne’s “hot top” — which sustains rapid heat loss even as the average temperature falls. Bregović’s winning essay emphasized that convection in the early cooling stage can keep accelerating the hot sample’s heat loss after it has caught up to the cold sample’s temperature.
Dissolved gases and impurities. Heating drives dissolved gases out of water, changing its composition slightly. Since dissolved substances shift freezing points and provide nucleation sites, the hot sample and the cold sample are not quite the same liquid anymore.
Supercooling. The current favorite. Water rarely freezes at exactly 0°C; it typically supercools a few degrees first, and the exact nucleation temperature varies unpredictably from sample to sample. If the cold sample supercools deeply while the hot sample nucleates promptly, the hot one wins — not because it cooled faster, but because it started freezing sooner.
Hydrogen-bond memory. The most exotic proposal came in a 2013 theoretical paper arguing that the effect is intrinsic to water itself: heating stretches and stores energy in water’s hydrogen bonds, and the bonds release that energy faster during cooling — a kind of molecular memory of having been hot. It is an elegant idea. It is also, like the others, unconfirmed as the general explanation.
The honest summary is the one the literature keeps converging on: the outcome depends on the container, the freezer, the water’s purity and history, the definition of “frozen,” and the luck of nucleation. Change any of these and the result can flip. That is why careful experimenters can get opposite answers and both be right about their own apparatus.
Why the viral version is misleading
Now return to the claim as it usually travels: hot water freezes faster than cold water, full stop — a general law, a kitchen hack, a paradox that defeats common sense.
That version is misleading in two specific ways. First, it presents as universal something that is conditional. The reproducible observations — Brownridge’s 28-for-28, the student replication, Osborne’s beakers — all occurred under particular, carefully described conditions. Under other equally careful conditions, Burridge and Linden found nothing at all. A phenomenon that appears or vanishes depending on the freezer’s temperature, the container’s shape, and the water’s dissolved-gas content is not a law of nature in the way the viral version implies. It is a sometimes-thing, and “hot water freezes faster than cold water” stated without qualification is the kind of sentence that makes a sometimes-thing sound like an always-thing.
Second, the viral version obscures what “freezes first” even means. In the successful experiments, the hot sample usually does not cool faster in the thermodynamic sense — it often nucleates ice sooner because of supercooling differences, or loses heat faster early on due to convection, or benefits from evaporation. The headline “hot beats cold” compresses all of that into apparent magic, and the magic is what goes viral. The reality — a subtle interplay of nucleation statistics and heat transfer, real but fragile — does not fit in a headline, so the headline drops the subtlety.
This is the classic anatomy of a misleading claim: a genuine observation, stripped of its conditions, polished into a rule. Mpemba’s own observation was real. The physics-teacher dismissal was wrong. But the leap from “in this freezer, with these beakers, the hot sample froze first” to “hot water freezes faster than cold water” is a leap the evidence does not support.
The Verdict
Misleading. Hot water has genuinely been observed to freeze before cold water under specific, reproducible conditions — Brownridge recorded it 28 times in 28 attempts, and the stochastic nature of ice nucleation gives the observation a sound physical basis. But as a general claim, it is false: under controlled conditions, hot water does not cool more quickly than cold water, as Burridge and Linden demonstrated in 2016, and simple thermodynamics — the hot sample must travel further to reach 0°C — says it cannot. The Mpemba effect is real the way a hole-in-one is real: it happens, it can be documented, but anyone who tells you it is the normal outcome is selling you something. The viral version drops the conditions that make the phenomenon possible, and without those conditions, the claim misleads.
There is a coda worth savoring. In 1963, a schoolboy who had just watched hot ice-cream mix beat cold mix to the freezer’s finish line asked his teacher why, and was told, in effect, to stop asking. He asked again anyway — of a visiting physicist, in front of a roomful of classmates who laughed. That question has now generated thousands of papers, a Royal Society of Chemistry prize, and an argument that has outlived both the teacher who dismissed it and, eventually, the certainty of everyone who thought they had settled it. The next time someone tells you that hot water freezes faster than cold water, you can tell them the full story: it is not true in general, it is true sometimes, and the difference between those two statements is the entire history of science in miniature.
