Tanzanian student Erasto Mpemba noticed something strange while making ice cream in the 1960s: his hot mixture froze before a cooler batch. That observation, later formalized with physicist Denis Osborne in a 1969 paper published in Physics Education, volume 4, issue 3, pages 172 to 175, launched decades of scientific argument. More than fifty years later, controlled experiments have produced contradictory results, theoretical models have extended the puzzle beyond water entirely, and researchers still cannot agree on a single explanation or even a shared definition of what “freezing faster” means.
Why the Mpemba effect still divides physicists
The core tension is deceptively simple. Fill two identical containers with water, one hot and one cold, place them in the same freezer, and ask which reaches solid ice first. Intuition says the cold sample should win every time. Yet scattered experiments over centuries, with observers as far back as Aristotle, Francis Bacon, and René Descartes, have reported the opposite outcome. The trouble is that positive results appear under some conditions and vanish under others, making replication inconsistent and fueling arguments over whether the effect is real or merely an artifact of experimental design.
One reason the debate persists is that researchers measure different things. Some track only the cooling phase, timing how long water takes to reach zero degrees Celsius. Others include the full freezing process, counting the minutes until the last liquid becomes solid. Because supercooling can delay nucleation unpredictably, the choice of endpoint changes which sample “wins.” A technical review by Monwhea Jeng mapped competing explanations, including evaporation, convection, dissolved gases, supercooling, and container effects, and argued that many published claims are not comparable precisely because they define “freezing” differently. Under one definition, a hot sample may appear to cool unusually quickly; under another, the same run may show no anomaly at all.
Evaporation is one of the most intuitive mechanisms proposed. Hot water can lose more mass as vapor, leaving less liquid to cool and freeze. That alone could shorten the time to reach the solid state, but only if evaporation is substantial and not offset by other factors. Convection currents are another candidate: temperature gradients in hotter water can drive internal circulation, potentially improving heat transfer to the container walls. Dissolved gases add further complexity. Heating water can drive gases out of solution, altering nucleation behavior and the way ice crystals form. Each of these mechanisms may operate at once, making it difficult to tease apart which is dominant in any given setup.
A hypothesis worth testing centers on container surfaces. Micro-roughness on vessel walls can seed ice nucleation at different rates, meaning that two containers that look identical may behave differently at the microscopic level. If researchers repeated trials across polished versus etched metal vessels while holding all other variables fixed, they could isolate whether surface texture, rather than bulk thermal properties, accounts for most positive Mpemba observations. No standardized multi-lab protocol using identical nucleation triggers has been published, leaving this variable uncontrolled in most existing data and raising the possibility that small differences in manufacturing or cleaning history might swing results from positive to negative.
Competing experiments and the definition problem
Henry Burridge and Paul Linden published a 2016 study that applied a strict cooling-only definition, measuring the time for water to drop from its starting temperature to zero degrees Celsius. Under those conditions, they found no consistent advantage for hotter starting temperatures. Their conclusion was direct: under their framework, the Mpemba effect did not appear. A subsequent technical rebuttal countered that restricting the measurement to the cooling phase alone misses the stages where the effect is most likely to emerge, particularly supercooling and nucleation. When those later stages are included, hot water can reach full solidification sooner than cold water in certain trials, especially when the colder sample supercools significantly below zero before finally freezing.
This disagreement is not just academic bookkeeping. It reflects a real gap in how experiments are designed. Raw temperature–time traces and container specifications from the original 1969 Mpemba and Osborne runs are not deposited in any public repository, so later researchers cannot fully reconstruct the conditions that produced the founding observation. Each new lab effectively starts from scratch, choosing its own vessels, volumes, cooling rates, and endpoints. Even subtle choices-such as whether lids are used, how thermometers are mounted, or how often samples are disturbed for measurement-can change airflow, nucleation sites, and heat transfer.
Recent work has pushed the question beyond water altogether. A team working with a colloidal system coupled to a thermal bath reported exponentially faster cooling from certain higher-energy starting states, a behavior they identified as a “strong” form of the Mpemba effect. In this system, the key variable was not phase change but the relaxation of particle distributions toward equilibrium. Separately, a theoretical paper in the Proceedings of the National Academy of Sciences formalized Markovian models of the Mpemba effect and its inverse, showing that anomalous relaxation behavior can arise in general nonequilibrium thermodynamic systems. In these models, the path a system takes through its state space can allow a hotter initial distribution to approach equilibrium faster than a cooler one, even when both are relaxing toward the same final temperature.
These findings suggest the phenomenon is not a quirk of water chemistry but a broader feature of how certain systems return to equilibrium from different starting states. Instead of asking only whether “hot water freezes faster than cold,” the theoretical work reframes the question: under what conditions does a system starting farther from equilibrium relax more quickly? The answer depends on details of the energy landscape and transition rates between microstates, not just on bulk temperature. That perspective may eventually feed back into experiments with water by clarifying which combinations of container, cooling environment, and initial preparation are most likely to produce anomalous freezing times.
Open questions and what to watch for next
Several gaps keep the debate alive. No group has yet published a replication protocol strict enough to control for evaporation losses, dissolved gas content, container surface texture, and nucleation seeding simultaneously across multiple independent labs. Without that kind of coordinated effort, positive and negative results will continue to accumulate in parallel, each defensible within its own setup but not directly comparable to the others. A well-designed consortium study could, in principle, map out the full parameter space and identify where a robust Mpemba effect appears, where it vanishes, and how sensitive it is to small perturbations.
The definitional split also remains unresolved. Whether “freezing faster” means reaching zero degrees sooner or completing the phase transition to solid ice sooner is not a trivial distinction, because the two definitions can produce opposite outcomes from the same data run. Until the research community converges on a shared standard, claims for and against the Mpemba effect will keep talking past each other. A pragmatic compromise might be to report both metrics-time to reach zero and time to full solidification-alongside detailed temperature curves and environmental conditions, allowing future analysts to re-interpret results under different criteria.
For anyone curious enough to try the experiment at home, the practical takeaway is that results will depend heavily on details most people never think about: the material and finish of the container, the mineral content of the water, the humidity inside the freezer, and how often the door is opened. A single trial that seems to show hot water freezing first does not settle the question, but it does echo the puzzle that has occupied scientists from Mpemba’s classroom to modern statistical physics. Whether the final verdict is that the Mpemba effect is a rare but real outcome under narrow conditions, or mainly a mirage created by uncontrolled variables, the search for an answer continues to illuminate how surprisingly subtle the act of freezing water can be.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
