Does Hot Water Freeze Faster Than Cold? The Mpemba Effect Explained (2026)

Put two identical cups of water in the freezer, one steaming hot and one at room temperature. Common sense says the cool one should freeze first, since the hot one has to lose all of its extra heat before it can even catch up. And yet for more than 2,000 years, people have insisted that hot water freezes faster.

The claim has a name, the Mpemba effect, after a Tanzanian schoolboy who refused to let a teacher laugh it off. It has a prize-winning explanation, a stack of rival theories, and a reputation as one of the most argued-over questions in kitchen physics. In 2026 alone, at least three new papers took a swing at it, and the latest, published on November 2, may finally explain why the result keeps showing up and then vanishing again.

The short answer: hot water can freeze before cold water, but only sometimes, under particular conditions, and mostly by luck. In careful tests, hot water takes longer to cool to 0 °C, as you would expect. Where it can win is the moment freezing actually starts, because water usually cools below 0 °C before ice forms, and the timing of that first ice crystal is partly random. In other systems, such as tiny particles in liquid and quantum bits, physicists have shown a reliable version of the effect.

A schoolboy, a batch of ice cream and an awkward question

In 1963, Erasto Mpemba was a 13-year-old student at Magamba Secondary School in what was then Tanganyika. During a cookery class, students were making ice cream by boiling milk with sugar, letting it cool and putting it in the freezer. Short on space and time, Mpemba put his mixture in while it was still hot. It froze before his classmates’ cooler batches.

His teachers told him he was confused. Mpemba was not convinced, and when physicist Denis Osborne of University College, Dar es Salaam, visited his later school in Iringa to give a lecture, the student stood up and asked him directly: if you put two similar containers of water, one at 35 °C and one at 100 °C, into a freezer, why does the one that started at 100 °C freeze first?

Osborne admitted he did not know, and he went back to his lab to try it. He found what Mpemba described. In 1969 the two published a short paper together in the journal Physics Education, with the cheerful title "Cool?" They had put 70 ml samples in 100 ml beakers in the ice box of an ordinary refrigerator, standing on foam. The time before freezing began was longest for water that started at about 25 °C and much shorter for water that started near 90 °C (Physics Education).

Mpemba went on to a career in wildlife management in Tanzania, but his name stuck to the puzzle. In January 2013 he was in London when the Royal Society of Chemistry announced the winner of a competition it had run for the best explanation of the effect. It drew 22,000 entries (Royal Society of Chemistry).

A short history of an old claim

Mpemba did not discover the idea so much as drag it back into the lab. The claim has a long trail behind it.

Around 350 BCE: In his Meteorology, Aristotle wrote that "the fact that the water has previously been warmed contributes to its freezing quickly: for so it cools sooner," and said many people set water in the sun first when they wanted to cool it fast (MIT Internet Classics Archive).

1620: Francis Bacon, in his Novum Organum, noted that slightly tepid water freezes more easily than water that is utterly cold.

1637: René Descartes repeated the claim, writing that water kept on a fire for a long time freezes faster than other water, and blamed it on particles boiling off.

1775: Scottish chemist Joseph Black tested the special case of previously boiled water against unboiled water and found the boiled water froze sooner, even when he controlled for evaporation. He also noticed that stirring the unboiled water made it freeze at the same time, a clue that the real story was about how water starts to freeze, not just how fast it cools (Philosophical Transactions of the Royal Society).

1963 to 1969: Mpemba’s ice cream, his question to Osborne, and their joint paper.

1995: David Auerbach showed that water often cooled well below 0 °C, to between −6 °C and −18 °C, before freezing, and that in some runs the hotter water began to freeze first.

2006: Writing in Physics World, Philip Ball listed how many things an experimenter has to control, from the container’s shape and material to dissolved gas, the freezer’s temperature and even frost on the shelf (Physics World).

2016: Henry Burridge and Paul Linden went back to the data and ran their own careful tests, and came away unconvinced (more below).

2020 to 2026: The effect moved into new territory, with reliable versions found in microscopic particles and in quantum systems, while water experiments kept turning up mixed results.

What the careful experiments found

The hardest part of the Mpemba effect is that "freezes faster" can mean three different things: reaching 0 °C first, starting to form ice first, or turning completely to ice first. Many disagreements come down to which of those an experiment measured.

2016: hot water does not cool faster

Burridge and Linden measured the simplest version, the time for water to cool to 0 °C. Their review found that the dramatic results in early experiments had not been repeated, and that small effects could come down to where the thermometer sat in the container. Their own tests showed no meaningful Mpemba effect, and their paper ended with a line that has been quoted ever since (Scientific Reports).

We conclude, somewhat sadly, that there is no evidence to support meaningful observations of the Mpemba effect. (Henry Burridge and Paul Linden, Scientific Reports, 2016)

James Brownridge had found much the same thing earlier: in his tightly controlled runs, which measured when freezing began, hot water sometimes won, but the container and how it touched its surroundings mattered as much as the starting temperature.

January 2026: 176 samples in a chest freezer

A study published on January 16, 2026, in RSC Applied Interfaces by Anna Janni, Santiago Botero Ampudia and E. Dan Dahlberg tried to work out why so many people still see the effect. The team put 35 ml water samples at different starting temperatures into a chest freezer running at about −14 °C to −17 °C, in lidded tins, open beakers, on foam and on a metal baking sheet, with both tap water and deionized water (RSC Applied Interfaces).

Across 176 samples in 25 runs, they counted 46 cases where a hotter sample reached 0 °C before a colder one and 58 cases where a hotter sample started freezing first. But there was no trend that made hot water cool faster. Instead, the cooling rate scattered from sample to sample, and the samples with the most scatter, tins sitting on a metal sheet, showed the effect most often. When the tins were boxed into foam to shield them from moving air, the effect almost disappeared. The authors’ conclusion: the "wins" came from random swirls of air in the freezer and from how far each sample happened to supercool.

Two popular explanations also came out badly. If evaporation were the cause, open beakers should have shown the effect more often than lidded tins; they showed it less. If dissolved gas were the cause, tap water should have shown it more than deionized water; it did not.

November 2026: freezing as a roll of the dice

The newest result came on November 2, 2026, in the journal Molecules. Andrei Klimov and Alexei Finkelstein of the Institute of Protein Research in Pushchino, Russia, placed four sealed 2 ml vials of water at +70 °C and four at +20 °C side by side, suspended on wires inside a stabilized chest freezer in a basement with vibration sources turned off, and recorded the moment each one began to freeze (Molecules).

Every hot vial reached 0 °C about 500 seconds after the cool ones, exactly as ordinary physics predicts. What happened next depended on the freezer. Pure water in a clean, sealed vial does not freeze at 0 °C; in their tests it started freezing at about −6 °C. When the freezer was only just colder than that, above about −7 °C, the wait for the first ice crystal became long and wildly variable, stretching to hours, and in roughly two of every five hot-versus-cool comparisons the hot vial froze first. When the freezer was colder than −8 °C, freezing started quickly and predictably, and the cool water froze first in 170 of 176 comparisons.

Their explanation is that the start of freezing is a random event. One tiny seed of ice has to form somewhere on the vial wall, and close to the freezing threshold the waiting time for that seed is scattered over a range far longer than the 500-second head start the cool water had. Sometimes the hot water simply gets lucky. No special "memory" in the water is required.

Why hot water can still win in a real kitchen

None of this means people who swear by the effect are imagining it. A home freezer is not a lab, and several ordinary things can tip the race toward the hot cup. Water at the other extreme has its own odd habits, like the Leidenfrost effect that sends drops skating across a hot pan, but the freezer version is all about timing.

Supercooling and the first ice crystal

Water often cools below 0 °C without freezing, then turns to ice all at once when a crystal finally forms. How far a sample supercools varies from cup to cup. If the cold cup happens to supercool further and wait longer, the hot cup can freeze first, which is the effect Klimov and Finkelstein say is down to chance near the threshold.

Frost and contact with the shelf

A hot container set on a frosty shelf can melt the frost under it and settle into better contact with the cold metal, while a cool container sits on an insulating layer of frost. Philip Ball pointed to this as one simple way a home test can be skewed.

Moving air and the container

In a freezer, heat leaves mostly through the air. The January 2026 study found that small differences in air currents around each sample changed cooling rates enough to flip the order, and that shielding the samples from moving air nearly erased the effect.

Evaporation and dissolved minerals

An uncovered hot cup loses some water to evaporation, leaving less to freeze, and boiling can drive minerals out of hard water. Both have been suggested as causes. Neither held up as the main one in the 2026 freezer study, but they can play a part in an open cup.

Where the Mpemba effect is real

The surprising twist is that physicists have found a solid, repeatable version of the effect, just not in a glass of water. In 2020, Avinash Kumar and John Bechhoefer of Simon Fraser University reported in Nature that a single microscopic bead jostling in water, in an energy landscape shaped with laser light, could cool exponentially faster when it started hotter, under the right conditions (Nature). The same group later showed an inverse version, where a colder system warms up faster than a less cold one.

Quantum physicists picked up the idea next. In 2024 two teams reported quantum Mpemba effects in trapped-ion experiments, and a Trinity College Dublin group worked out the thermodynamics of how a quantum system started further from equilibrium can settle faster (Trinity College Dublin). In January 2025, researchers in China reported a "quantum strong Mpemba effect" in a single trapped ion in Nature Communications.

The idea is now being put to work. A paper published on July 1, 2026, in the Journal of Physics A showed a way to use a Mpemba-style shortcut to reset qubits, the working parts of a quantum computer, up to 50% faster than letting them relax on their own, and tested it on an IQM superconducting quantum processor (Journal of Physics A). In these systems, there is no freezing at all, and no random ice seed to wait for; the speedup comes from which way the system starts out, not from luck.

The Mpemba effect: what we know and what is still unclear

What we know

Hot water does not cool to 0 °C faster than cold water in controlled tests; Burridge and Linden, Brownridge and the 2026 studies all agree on that. Hot water can begin to freeze first in some runs, mainly because water supercools and the timing of the first ice crystal varies. Containers, air currents and frost matter a great deal. And outside of water, in tiny particles and quantum systems, a reliable Mpemba effect has been measured and is being used.

What is still unclear

Whether there is any mechanism special to water at all is still argued over. Klimov and Finkelstein say no, at least for pure water in sealed vials; other groups have pointed to dissolved substances, hydrogen bonding or bubbles. The papers also do not all measure the same thing, so headlines that say the effect has been "proven" or "debunked" usually mean one definition. The new 2026 results have not yet been independently repeated.

Questions people ask

Should I use hot water to make ice cubes faster?

No. For ice cubes in a typical freezer, starting with cold water is the better bet. The 2026 chest-freezer and vial studies both found that, most of the time and especially in a cold freezer, cooler water freezes first. Hot tap water can also carry more dissolved metals from pipes, which is a reason to avoid drinking or freezing it anyway.

Why did my hot water freeze first, then?

It probably did, that one time. Small differences in where each cup sat, the frost under it, the air flowing past it and how long each cup supercooled can easily flip the result. Run the test several times and the winner tends to change.

Can I test it at home?

Yes, but the details make or break it. Use identical containers with the same amount of water, put lids on them, place them side by side on the same kind of surface, and decide in advance what "frozen" means: the first ice, or solid all the way through. A cheap thermometer helps. Then repeat it many times and count, rather than trusting a single race.

Is it fair to name it after Mpemba when Aristotle wrote about it first?

Most physicists think so. Aristotle, Bacon and Descartes reported the claim, but Mpemba and Osborne were the first to test it in a modern, published experiment and set off the research that followed. The name also honors a teenager who kept asking a question his teachers dismissed.

The bottom line

Hot water does not have a hidden turbo button. In a fair race it takes longer to get cold, and in a cold enough freezer it almost always loses. But the moment water turns to ice is a surprisingly chancy event, and in the gap between cooling and freezing, the hot cup sometimes slips ahead. More than 60 years after a schoolboy asked an awkward question, the answer turns out to be both simpler and stranger than anyone expected: sometimes hot water freezes first, and sometimes it is just luck.

Related on aglena: The Coin That Screams Toward Infinity: The Strange Physics of Euler’s Disk, The Ice King: How Frederic Tudor Sold New England’s Frozen Ponds to the Tropics

Further reading