Whether hot water freezes faster is still not properly settled

Most explanations of hot water freezing faster than cold stop at the point where it starts to matter. This one carries on.

Where the claim comes from

Ancient and early modern writers on natural philosophy mention that warmed water sometimes freezes sooner, usually as an observation from practical life. The modern version began when a schoolboy in Tanzania noticed that a hot ice-cream mixture froze before a cold one and asked why.

His teacher told him he was mistaken, which is a reasonable thing to say, and the boy persisted, which is a better one. A visiting physicist took the question seriously, ran experiments with him and published the result, and the effect acquired the boy's name. The origin of the modern claim is therefore unusually well documented, and involves a student refusing to accept a brush-off.

Why it sounds impossible

Cooling water from a higher temperature means passing through every intermediate temperature the cold sample started at. If nothing else differs, the hot sample should arrive at the cold sample's starting condition and then be behind for the rest of the process.

Trace it back and the argument only holds if the water's state is fully described by its temperature, which for a real container of water it is not. Heating changes dissolved gas content, may drive off some water as vapour, and sets up convection currents that persist for a while. So the paradox dissolves in principle, and the remaining question is whether any of these differences is large enough to matter in practice.

The proposed mechanisms

Evaporation reduces the mass of the hot sample, and a smaller mass has less heat to lose, though the effect is usually modest in a covered container. Dissolved gases are driven out by heating, and their absence may alter how the water behaves as it approaches freezing. Convection driven by the temperature difference can distribute heat differently and change how the sample loses energy at its surface.

Tested properly, supercooling is a serious complication, since water often stays liquid below its freezing point until something triggers crystallisation. If two samples supercool to different extents, the one that freezes first may simply have found a nucleation site sooner, which is close to chance.

The replication record

Attempts to reproduce the effect under controlled conditions have produced a mixed picture, with some studies reporting it and others finding nothing. A careful set of experiments published in the past decade, controlling the container, the sensor placement and the definition of freezing, failed to observe the effect.

Where the story actually begins, much depends on what counts as frozen, since reaching zero degrees, forming the first ice and freezing solid are three different events. Different definitions can turn the same experimental run into a confirmation or a refutation, which is a serious methodological problem.

The field does not have an agreed protocol, which is exactly the condition in which a contested effect can persist indefinitely.

Why it is hard to test

Freezing water involves a container, a freezer with its own temperature cycles, air movement, and the position of whatever is measuring the temperature. Freezers are not steady environments, and the compressor cycling on and off introduces variation larger than the effect being sought.

Checked against the record, supercooling is stochastic, so any single comparison is close to meaningless and large numbers of runs are needed. Very few laboratories have an incentive to run hundreds of trials on a question with no application attached to it. This is a good example of a simple-sounding question that is genuinely expensive to answer properly.

Printing a correction has a poor record of taking a myth out of circulation.

What to say about it

The honest position is that the effect is reported under some conditions, that no mechanism is agreed, and that careful replications disagree. That is less satisfying than either confident answer, and it is what the evidence currently supports. It is a useful case because it shows that a question can be simple, old, famous and still open.

Somewhere in the retelling, it also shows how much of an experimental result lives in the definitions, which is not a lesson that survives being summarised. In a domestic freezer, filling the tray with cold water remains the sensible default, since the disputed effect is not something you can rely on.

The takeaway

A schoolboy asked a question in the 1960s and physicists still cannot agree on the answer, which is the most interesting part.

Believing it was ordinary. Continuing to is the avoidable part.