Sunday, March 21, 2010

A Meditation on Two Papers on the Mpemba Effect

I came across two excellent, fascinating, and very different papers this morning, about the Mpemba Effect: the observation that hot liquids often take less time to freeze than cold liquids, an apparent contradiction of Newton’s laws of thermodynamics. I love the story of the Mpemba Effect, it’s exactly the kind of story that scientists love to tell: scrappy upstart notices odd experimental findings, receives just enough scoffing to make for a good story (but not so much that he gives up -- he is made of sterner stuff than that!), then finally proves himself and vindicates his findings. Hooray!

OK, he wasn’t really the first person to notice it. Aristotle (in Meteorology, Book 1, near the end of Part 12) noticed it... but then, Aristotle was not much of an experimentalist: he used to claim that women had fewer teeth than men, for example. Despite having had three wives, it never occurred to him to, y’know, count. (On the other hand, if Aristotle asked to count my teeth, I’d say no.)

I was struck, in reading the two papers I’m about to describe, both by how well-crafted they are and how very different they are from each other. They complement each other in important ways, and I highly recommend reading (or at least skimming) both, as they are both freely available through Cornell’s arXiv, which is generally my go-to place for Interesting Science Crud and an excellent resource for would-be science-fiction writers.

The first paper (actually, the second one I read, but you should read it first) is an excellent historical overview of the problem, by Monwhea Jeng. It does a very able job of answering the question, “What is this effect, why is it interesting, and why should it be studied?” as well as some possibilities that others have raised for explanations. Bottom line: this problem is real, weird, and worth pursuing. (Jeng is a very able writer, too, and paints a much more interesting portrait of Mpemba’s experience than does Wikipedia)

I did take issue with one statement, though:
What is interesting about the Mpemba effect is that unlike the examples commonly given in science text- books, where theory and experiment march hand-in- hand, always leading to further progress, here theory (rightly or wrongly) prevents acceptance of experiment.
Now, this paragraph (which I have admittedly scrubbed of context) is more subtle than it looks: my first reaction was to shake my head and say that Jeng has it exactly backwards, as Gödel, Darwin, or many others would attest. But the bit about the science textbooks gives me pause here. I remember my textbooks as playing up the scientist-as-lone-hero aspect. Has that changed? Am I misremembering?

The second paper I read this morning, by James Brownridge of SUNY Binghamton, is an attempt to bring together all the possible experimental conditions that give rise to this apparent effect. This one is a very thorough experimental paper that comes to an explanation that I find very convincing. (Hint: the definition of the Mpemba effect is that, under the same experimental conditions, a quantity of hot liquid freezes faster than the same quantity of otherwise identical cold liquid. Ask yourself what it really means to have the same experimental conditions) Brownridge goes to great lengths to examine the problem from many different aspects, but also maintains something of a narrative: this was done this because of this, and then this and that as a result.

The contrast between the two papers is, to me at least, striking. I was tempted at the outset to put more value in the Brownridge paper with its detailed experiments, charts, and explanations. But I was corrected by none other than Brownridge, who holds up Jeng’s paper in the first paragraph as helpful and useful.

So, I backpedaled, and thought about it. There are two important pieces here: the problem and the solution. Often in the papers I read (and write) the two pieces come from the same author, who frames the problem, describes the procedures, and presents the solution all in one paper. It’s difficult under those conditions to avoid warping the definition of the problem to make the solution look better: after all, you’re not just persuading the reader that you’ve done useful science, but the publishers and reviewers of that paper, and to some extent convincing yourself and your teammates. I think that using another person’s paper as a problem definition can help keep you honest -- it helps outline the work a little better, you can’t rephrase it or reframe it in convenient ways.

Oh, and here’s another way to improve the discussion of science.

No comments:

Post a Comment