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

A Tanzanian Student Noticed It Making Ice Cream in 1963 — Scientists Still Cannot Agree Whether It Is Real

In 1963, Erasto Mpemba, a secondary school student in Iringa, Tanzania, was making ice cream during a school cooking class. Short on freezer space, he put his ice cream mixture into the freezer while it was still hot, without waiting for it to cool. He noticed that his hot mixture froze faster than classmates' cooled mixtures. When he raised the observation with his teacher, he was dismissed. Several years later, Mpemba discussed the phenomenon with Denis Osborne, a physicist from the University of Dar es Salaam visiting his school. Osborne took the question seriously and confirmed it experimentally. Their joint paper, published in Physics Education in 1969, attached Mpemba's name to the phenomenon for the scientific record. What followed was six decades of contested experiments, proposed mechanisms, and unresolved disagreement among physicists and chemists — with the answer still genuinely unclear.

Defining the Effect Precisely

The Mpemba effect, as defined in the scientific literature, is the claim that under some conditions, a sample of water at a higher initial temperature will freeze in less time than an otherwise identical sample at a lower initial temperature, when both are placed in the same freezer. Critically, "freeze" can be defined in different ways — reaching 0°C, becoming fully solid, or reaching some other reference point — and the experimental outcome is sensitive to which definition is used. This definitional ambiguity is a significant source of the controversy.

The claim seems counterintuitive. Naively, if Sample A starts at 90°C and Sample B starts at 20°C, Sample B has less thermal energy to lose before freezing, so B should freeze first. The Mpemba effect says that sometimes A freezes first anyway. Not always. Not reliably. Sometimes.

Proposed Mechanisms

Numerous physical mechanisms have been proposed to explain the Mpemba effect. None has achieved consensus as the sole or primary explanation, and different mechanisms may account for the effect under different experimental conditions.

The 2016 Royal Society of Chemistry Competition

In 2012, the Royal Society of Chemistry ran a public competition asking for the best scientific explanation of the Mpemba effect, attracting over 22,000 entries. The winner, physicist Nikola Bregović from the University of Zagreb, proposed an explanation centered on convection and supercooling differential. The competition raised the phenomenon's scientific profile but did not resolve the underlying physics — the prize entry was an explanatory hypothesis, not experimental confirmation.

The 2016 Negative Result

A landmark 2016 study by Henry Burridge and Paul Linden at the University of Cambridge, published in Scientific Reports, systematically examined the experimental literature and found that many positive Mpemba effect reports were methodologically flawed — using inconsistent container geometries, variable freezer conditions, or inadequate controls. When Burridge and Linden performed carefully controlled experiments, they found no reliable Mpemba effect under the controlled conditions. Their conclusion: the effect, when observed, is likely the result of uncontrolled variables rather than a fundamental physical phenomenon. The scientific community has not reached consensus on this conclusion either.

The 2023 Molecular Simulation Study

Research published in the Journal of Chemical Physics in 2023 by Zhiyuan Niu and colleagues used molecular dynamics simulations to examine water cooling at the molecular level. Their simulations showed conditions under which hot water samples cooled faster — attributing the result to differences in the structure of hydrogen bond networks at different temperatures. The simulations supported the possibility of a real Mpemba effect but at nanoscale, which may not scale to the macroscopic experiments typically used to test the phenomenon. This highlights one of the persistent challenges: the mechanisms plausible at the molecular scale are difficult to connect cleanly to observable macroscopic freezing times.

Experimental Challenges and Why Reproducibility Is Difficult

The Mpemba effect is notoriously difficult to reproduce reliably. The following variables all affect experimental outcomes and are often inadequately controlled in published studies.

• Container material and geometry: Glass, plastic, and metal containers transmit heat at different rates; container shape affects convection patterns
• Freezer airflow and shelf position: Placement within a freezer significantly affects local heat exchange; consumer freezers have highly variable internal temperature distributions
• Water source: Dissolved mineral content, dissolved gases, and prior thermal history of the water all affect cooling behavior and nucleation
• Definition of "freezing": Surface freezing, thermocouple reading at 0°C, and complete solidification produce different results from the same experiment
• Repetitions: Many published positive results used small sample sizes; the effect, if real, appears to be stochastic rather than deterministic

Historical Antecedents

Mpemba was not the first to notice the phenomenon. Francis Bacon referenced it in his Novum Organum (1620): "Water a little warm is more easily frozen than quite cold." René Descartes mentioned a similar observation in his 1637 work Les Météores. Aristotle arguably described the phenomenon in the Meteorologica (350 BCE), noting that "water that has been heated contributes to its freezing more quickly." The ancients observed something. Whether that something is a robust physical phenomenon or a collection of confounded experimental artifacts remains the question researchers continue to investigate more than 2,400 years after Aristotle first wrote it down.