I have a piece in the Guardian online on the Mpemba effect and the RSC’s £1000 prize for explaining it. The article is largely unchanged from what I wrote, but here it is anyway.
The aim here was to stimulate suggestions from readers of how this thing can be explained – or even if there’s a real effect to be explained, though few seem to question that. (One does so with amusing literalism, thinking it implies that hot water will always freeze first whatever the temperature difference. All the same, this reinforces Charles Knight’s point that the phenomenon is too ill defined.) I like too the cute popular notion of “heat loss momentum” – check out Newton’s cooling law, please.
But I’m certainly not going to mock the many confused or just plain wrong suggestions put forward, since the whole point of the exercise is to get people engaged, not to laugh at their errors. However, I can’t help being struck by the inevitable one or two who say, apparently in all seriousness, that the answer is just obvious and everyone but them has been too stupid so far to see it. One chap dispenses with all of the additional ‘mysteries’ in the article this way too. Why can all arms of a snowflake sometimes be identical? “The symmetry comes from the initial nucleation of the crystal. It starts symmetrically and keeps growing symmetrically. And computer simulations have shown this.” I can only assume he/she (probably he) has seen some simulated flakes and failed to read the warning that symmetry was imposed on all six arms. He certainly didn’t think it worth bothering to check out the link to Ken Libbrecht’s page, which makes it clear that (as I said in the piece) the side-branches are, according to the standard theory of dendritic growth, amplified randomness. So the entire form of any given flake is somehow inherent in its initial nucleus? Please. I couldn’t help smiling too at the apparent belief of some readers that the Brazil-nut effect was actually discovered in muesli (leading to a discussion on how muesli gets packaged). Anyway, the comments thread provides a nice little cross-section of how folk think about science. And, I think, somewhat encouraging at that, despite the misconceptions.
If you can explain, before the end of July, why hot water freezes faster than cold, you could bag £1000. That’s what the Royal Society of Chemistry (RSC) is offering for “the most creative explanation” of this phenomenon, known as the Mpemba effect. They say that submissions should be “eye-catching, arresting and scientifically sound”, and may use any media, including film.
At the end of the month the problem will also be put to an international summer school for postgraduate science students called Hermes 2012, convened at Cumberland Lodge in Windsor Great Park to present research in materials science and imbue the participants with skills in science communication. The event, organized by Imperial College and sponsored by the RSC, is timed to coincide with the opening of the Olympic Games as a kind of scientific Olympiad. A presentation of the top entries to the RSC’s competition, alongside the efforts of the meeting attendees, will form a highlight of the event on 30 July.
All good fun – except that the Mpemba effect seems at first encounter to be scientific nonsense. Let’s have that again: “why hot water freezes faster than cold”. How can that be? In order to freeze, hot water has to lose more heat than cold, so why would that happen faster? Even if the cooling of hot water somehow catches up with that of the colder water, why should it then overtake, if the two have at that point the same temperature?
Yet this effect has been attested since antiquity. Aristotle mentions it, as do two of the fathers of modern science, Francis Bacon and René Descartes in the seventeenth century. The effect is today named after a Tanzanian schoolboy, Erasto Mpemba, who was set the project of making ice cream from milk in the 1960s. The pupils were supposed to boil their milk, let it cool, then put it in the fridge to freeze. But Mpemba worried about losing his space in the fridge, and so put in the milk while it was still hot. It froze faster than the others.
When Mpemba learnt a few years later that this seemed to contradict the theory of heat transfer devised by Isaac Newton, he recalled his experiment and asked his teacher to explain it – only to receive a mocking reply. Undeterred, he carried out his own experiments, and asked a visiting university professor from Dar es Salaam, D. G. Osborne, what was going on. Osborne was more open-minded – he asked his technician to repeat the experiment, and found the same result. In 1969 Osborne published the result in a physics education journal. Coincidentally, that same year a physicist in Canada described the same result, saying that it was already folk wisdom in Canada that a car should be washed with cold water in winter, because hot water froze more quickly.
Yet no one really knows if the Mpemba effect is real. You’d think it should be easy to check, but it isn’t. Ice specialist Charles Knight of the National Center for Atmospheric Research in Boulder, Colorado, says that the claim that “hot water freezes faster than cold” is so ill-defined that it’s virtually meaningless. Does it mean when ice first starts to appear, or when the last bit of water is frozen? Both are hard to observe in any case. And there are so many things you could vary: the amount of water, the shape of the containers, the initial temperature difference, the rate of cooling… Do you use tap water, distilled water, de-aerated water, filtered water? Freezing is notoriously capricious: it can be triggered by tiny scratches on the sides of the flask or suspended dust in the liquid, so it’s almost impossible to make truly identical samples differing only in their starting temperature. For this reason, even two samples starting at the same temperature typically freeze at different times. If such ‘seeding’ sites are excluded, water can be ‘supercooled’ well below freezing point without turning to ice – but here experiments are conflicting. Some find that initially hotter water can be supercooled further, others that it can be supercooled less before it freezes.
There is one trivial explanation for Mpemba’s observations. Hot water would evaporate faster, so if there was no top on the flasks then there could have been less liquid left to freeze – so it would happen faster. Tiny gas bubbles in solution could also act as seeds for ice crystals to form – and hot water holds less dissolved gas than cold.
All this means that a single experiment won’t tell you much – you’ll probably have to do lots, with many different conditions, to figure out what’s important and what isn’t. And you’ve only got a month, so get cracking.
Other mysteries to solve at home:
1. Why do the Brazil nuts gather at the top of the muesli? There’s no complete consensus on the cause of the so-called Brazil nut effect, but current explanations include:
- shaken grains in a tall box circulate like convection currents while the big bits get trapped at the top, excluded from the narrow descending current at the sides
- little landslides in the void that opens up temporarily under a big grain as it is shaken upwards ratchet it ever higher
- it's all to do with the effect of air between the grains
The problem is made harder by the fact that, under some conditions, the big grains can sink to the bottom instead – the ‘reverse Brazil nut effect’.
2. Does the water in a bathtub spiral down the plughole in opposite directions in the Northern and Southern Hemisphere? Cyclones rotate counterclockwise in the north and clockwise in the south, a consequence of the Earth’s rotation called the Coriolis effect. But is the effect too weak to govern a plughole vortex? In 1962 an American engineer named Ascher Shapiro claimed that he consistently observed counterclockwise plughole vortices in his lab, but this result has never been verified. The problem is that it’s really hard to rid a bathtub of water of any residual currents that could bias the outcome.
3. Why are all six arms of a snowflake sometimes (but not always) identical? How does one arm know what the other is doing? The standard theory of snowflake formation explains the ornate branching patterns as amplifications of random bumps on the sides of needle-like ice crystals. But if they’re random, how can one arm look like another? One suggestion is that they listen to one another: acoustic vibrations in the ice crystal set up standing-wave patterns that dictate the shape. But this doesn’t seem to work. Most snowflakes aren’t actually as symmetrical as is often supposed – but the fact that some are is still unexplained.