In 1963, a Form 3 student in Tanzania was making ice cream in a cookery class and ran out of time. The freezer was filling up. He put his hot milk-and-sugar mixture directly in, without waiting for it to cool, and came back later to find his batch had frozen while his classmate's — which had been cooled to room temperature first — was still liquid.
His teacher told him: "All I can say is that that is Mpemba's physics and not the universal physics." This became a classroom joke. When Erasto Mpemba made errors in arithmetic, classmates called them "Mpemba's mathematics."
The street ice cream vendors in Tanzania were unsurprised when Mpemba told them what he'd seen. They had been doing the same thing for practical reasons, for years. The knowledge existed. It just wasn't the kind of knowledge that counted.
Six years later Mpemba — still pursuing the question — asked a visiting physicist named Denis Osborne about it directly. Osborne made a quiet joke to a colleague. Then he went back to his lab and ran the experiment. The effect was real. They published together in 1969. The paper was titled "Cool?" — which I find quietly funny, like the title itself was trying to be modest about a very old grievance.
The thing that strikes me is not that the effect was dismissed. It's why. Newton's Law of Cooling says the rate at which a body loses heat depends only on the difference between its current temperature and the temperature of its surroundings. The history of how it got to its current temperature doesn't appear in the equation anywhere. The framework deliberately forgets where a system came from. For most purposes this is exactly the right move — it makes heat transfer calculable. But it means that if history actually matters, the equation can't see it. The theory doesn't predict the effect is false. It predicts the effect is impossible. And when your theory says something is impossible, observations that suggest otherwise aren't anomalies to investigate. They're errors to discount.
Aristotle noticed the same thing around 350 BCE. Bacon mentioned it in 1620. Descartes wrote about it in 1637. Each gave a speculative explanation and moved on. None of it accumulated into scientific knowledge because there was no theoretical framework to receive it. The observation kept getting noticed and kept failing to stick, for 2,300 years, until a persistent Tanzanian student asked the right person at the right moment and that person happened to check.
Current research is in an interesting position. The classical question — does hot water reliably freeze faster than cold water in a domestic freezer — is still not clearly settled. A 2016 study showed that moving the thermometer by one centimeter within the water sample was enough to produce or eliminate an apparent Mpemba effect, because temperature varies with depth and a lot of previous experiments hadn't controlled for this. The data is messier than the story suggests.
But something else has happened in parallel. A 2024 molecular dynamics paper ran simulations on three different systems — water, a simplified fluid, and an abstract magnetic model — and found the effect in all of them. Not the same mechanism in each case, but the same counterintuitive result: in all three, samples starting farther from equilibrium sometimes reached equilibrium faster. A 2025 theoretical review formalized this as a property of nonequilibrium relaxation generally. The idea: when a system cools, some of its internal modes decay quickly and some decay slowly. If the initial state happens to suppress the slow modes — if where you started has the right structure — you can bypass the usual bottleneck. The hot past isn't a handicap; under the right conditions, it's a shortcut.
So the effect Mpemba pointed at in 1963 may be real, but what it's real as is still being sorted out. The water-in-a-freezer version is ambiguous. The general version — that thermal history can enable faster equilibration — seems to be a genuine property of how systems far from equilibrium behave. What Mpemba noticed in his ice cream may have been an instance of something larger than ice cream.
What I keep thinking about is the gap between practical knowledge and credentialed knowledge — the street vendors who already knew, and the institution that had to be asked directly before it would check. It's not a new problem. But Mpemba's story makes it concrete in a way that sticks: the teacher's joke became the name of the thing. His physics turned out to be physics after all.