The Mpemba Effect (or ‘an apology to Morgan’)

Last Friday night (yes, this is how we spend our weekends) Morgan asked Robin and me a question: is it true that hot water will freeze faster than cold water? Apparently, he had gotten into an argument with someone who pointed to the obvious absurdity of the statement, and Morgan couldn’t remember exactly why this is true, but Materials_lab1 did remember reading somewhere that it is in fact true. In a manner superior and smug enough to indicate my superior knowledge of physics, I informed Morgan that he was wrong, and as a reference, told him that I remember Cecil Adams tackling this silly myth in The Straight Dope column some years ago. Sure enough, a quick search of The Straight Dope archives revealed this:

Dear Cecil:

I have a friend who insists that filling an ice cube tray with warm water will cause the cubes to form more quickly than they would if you started with cold water. He said it had something to do with the air circulation around the trays being affected by the temperature.

Not knowing much about frigidity myself, but being contrary, not to mention skeptical, by nature, I expressed doubt. Cecil, was I right, or is there indeed some basis in fact for this foolishness? –Mary M.Q.C., Santa Barbarba, California

Cecil replies:

You were smart to let me handle this, Mary. God knows what would happen if you tried to experiment with ice cubes on your own.

Needless to say, I conducted my research in the calm and systematic manner that has long been the trademark of Straight Dope Labs. First, I finished off a half a pint of Haagen-Dazs I found in the fridge, in order to keep my brain supplied with vital nutrients.

Then I carefully measured a whole passel of water into the Straight Dope tea kettle and boiled it for about five minutes. This was so I could compare the freezing rate of boiled H20 with that of regular hot water from the tap. (Somehow I had the idea that water that had been boiled would freeze faster.)

Finally I put equal quantities of each type into trays in the freezer, checked the temp (125 degrees Fahrenheit all around), and sat back to wait, timing the process with my brand new Swatch watch, whose precision and smart styling have made it the number one choice of scientists the world over.

I subsequently did the same with two trays of cold water, which had been chilled down to a starting temperature of 38 degrees.

The results? The cold water froze about 10 or 15 minutes faster than the hot water, and there was no detectable difference between the boiled water and the other kind. Another old wives’ tale thus emphatically bites the dust. Science marches on.

However, a reader wrote in to say that Scientific American had an article about why hot water freezes faster than cold, and old Unca Cecil had to revise his conclusion:

I know it must unnerve you to find that a supposedly infallible source of wisdom can make mistakes, so let me hasten to reassure you: Scientific American did not screw up. My results and theirs (specifically, those of Jearl Walker, author of SA’s “Amateur Scientist” column) are consistent–we were just working in different temperature ranges.

I found that cold water (38 degrees Fahrenheit) froze faster than hot water out of the tap (125 degrees F). I chose these two temperatures because (1) they were pretty much what the average amateur ice-cube maker would have readily available and (2) I couldn’t find a mercury thermometer that went higher than 125 degrees.

Jearl, who is not afflicted with penny-pinching editors like some of the rest of us, was able to get his mitts on a thermocouple that could measure as high as the boiling point, 212 degrees F. He found that water heated to, say, 195 degrees would freeze three to ten minutes faster than water at 140-175 degrees. (There were differences depending on how much water was used, where the thermocouple was placed, and so on.)

Jearl suggested that the most likely explanation for this was evaporation: when water cools down from near boiling to the freezing point, as much as 16 percent evaporates away, compared to 7 percent for water at 160 degrees. The smaller the amount of water, of course, the faster it freezes.

In addition, the water vapor carries away a certain amount of heat. To test this theory, Jearl covered his lab beaker with Saran Wrap to prevent water vapor from escaping. The freezing rate difference was greatly diminished. Conceivably convection (motion within the water) also plays a role.

Fascinating as all this no doubt is, all it basically proves is that very hot water freezes more slowly than very VERY hot water. The ordinary fumbler in the fridge, on the other hand, is dealing with temps more like the ones I was measuring, in which case cold freezes faster than hot. I rest my case.

You can see the Straight Dope page here. However, this is not the whole story either. Apparently the phenomenon in question is known as the Mpemba Effect after a Tanzanian high school student who discovered it in 1969. There is more to the story, and the possible mechanisms behind this anomalous effect here, here, here, here, and from Scientific American’s “Ask the Experts” column here.

My apologies to Morgan. (But I did double check. Can I get partial credit?)