The Buttered Toast Experiments
and other tests of everyday sayings
summer 1999 · conducted by Nathan, age 17

CONTENTS
1. Experiment 1: Buttered Toast
2. Experiment 2: The Watched Pot
3. Experiment 3: Grocery Lines
4. Experiments I Was Not Allowed to Do
5. Planned for Winter Break (separate page)

Experiment 1: Does Toast Really Land Butter-Side Down?

Hypothesis. Toast that falls off a table will land butter-side down more than half the time. My first guess at why was that the butter makes that side heavier.

Method.

[Mom's note: Two hundred slices of buttered toast. On my kitchen floor. In July. I would like that in the Method.]

Results.

Control: "Down" means X-side down. Total slices: 250, if you count the control, which Mom does.

So why does it happen? It isn't the butter. The unbuttered control slices landed X-side down 72% of the time, almost exactly the same as buttered toast from the same table. A level teaspoon of butter weighs only about 5 grams, and a slice of toast weighs a lot more than that, so the butter barely changes anything. My first hypothesis about why was wrong.

What actually happens is rotation. When a slice slides off the edge of a table, it doesn't fall flat. The part hanging over the edge tips down first while the rest is still resting on the table, so the slice is already spinning by the time it leaves. From normal table height there is only enough time for it to turn about half a turn before it hits the floor, and half a turn takes a slice that started butter-side up and lands it butter-side down. My physics teacher, Mr. Legrand, showed me an article about this by a British scientist named Robert Matthews in the European Journal of Physics in 1995, which says that to get a full turn, so the toast lands butter-side up again, you would need a table about 3 meters (10 feet) high.

That explains the counter. From 36 inches the toast has a little more time to turn. Most slices still only got around about half a turn, but the ones that were spinning fastest got far enough to come back around to butter-side up, so more of them made it. The higher drop changed the odds, just not by enough to save you.

Conclusion. Toast that slides off a table does land butter-side down most of the time (74% in my test). It's not bad luck and it's not the butter. It's the height of the table. Hypothesis supported, but my explanation was wrong, which Mr. Legrand says is still a success because now I know.

Sources of Error.

• Push speed. On the second day my brother Shawn did 12 of the pushes while I wrote, and he pushed faster. Fast slices left the table before they had tipped very far, turned less, and landed butter-up more often. I threw out his 12 and redid them myself. This actually agrees with the rotation explanation.
• Toast curl. Some slices curled up a little at the edges as they cooled, and curled slices sometimes bounced.
• Bounces. I counted the side that ended up facing the floor when it stopped moving. About 1 in 15 slices bounced, and a bounce can flip it.
• Butter amount. A "level teaspoon" depends on how soft the butter is. It was July.
• The dog. Gus got to three slices before I could record them. I redid those too.

[Mom's note: The butter-up slices were eaten by Nathan and Shawn. The butter-down slices went to the birds in the backyard, who have never been fatter. The floor is still not right.]

Experiment 2: Does a Watched Pot Boil?

Hypothesis. A watched pot boils in the same amount of time as a pot nobody is watching, because the water doesn't know.

Method. Same saucepan every time, filled with 2 quarts of cold tap water (starting temperature checked with a cooking thermometer, 61°F every time, within a degree). Same burner on our electric stove, set to high, no lid. "Boiling" means a full rolling boil, with big bubbles breaking all over the surface that don't stop when you stir. Five watched trials: I sat on a stool and looked at the pot the whole time, starting the stopwatch when I turned on the burner and stopping it at the boil. Five unwatched trials: I turned on the burner, started the stopwatch, and left the kitchen. The family camcorder recorded the pot, and I timed the boil from the tape afterward. Between trials the burner cooled for 30 minutes so every trial started cold.

Results.

* Watched trials only. Before looking at the stopwatch, I wrote down how long I thought it had been.

Conclusion. A watched pot boils. The average times were 1.2 seconds apart, which is less than the error in my thumb on the stopwatch. Hypothesis supported. However, the "felt like" column shows that a watched pot feels like it takes almost twice as long, which I think is where the saying comes from. The saying is not about water. It is about people.

Sources of Error.

• The camcorder may count as watching. I asked Mr. Legrand and he said that is a philosophy question and he teaches physics.
• Stopwatch reaction time, maybe a quarter of a second each way.
• Deciding exactly when a boil is "rolling" is a judgment call. I tried to use the same judgment every time.
• On trial 4 (watched), I was very bored and may have blinked a lot.

[Mom's note: This is the only one of the three experiments I endorse. Ten pots of water, no mess, and he made spaghetti with the last one.]

Experiment 3: Which Grocery Line Is Fastest?

Hypothesis. The checkout line with the fewest people in it is the fastest.

Method. I asked the manager of the grocery store where my mom shops for permission first. (He asked if I worked for the other grocery store. I said no, I work for science.) On four Saturday mornings in August, 10 a.m. to noon, I stood at the end of the checkout lanes with a clipboard and a stopwatch. When a customer got in line, I wrote down how many people were ahead of them, about how many items were ahead of them (counting carts as well as I could), and whether it was the express lane. Then I timed them from getting in line until they finished paying, and wrote down anything that slowed things down. I timed 60 customers. Then, for each customer, I looked at whether they had picked the shortest line, the line with the fewest items, or the express lane, and averaged the waits.

Results.

Conclusion. Hypothesis not supported. Counting people is not as good as counting items, because one person with a full cart takes longer than three people with a few things each. The express lane was fast per person but usually had the longest line. But the biggest delays, price checks and checks, can't be predicted by looking at a line, and one price check can wipe out any strategy. Best advice from my data: pick the line with the fewest items, and if the person in front of you is holding a checkbook, think about moving.

Sources of Error.

• I couldn't always see every item in every cart.
• Only 60 customers, only Saturday mornings, only one store.
• One lane had a bagger and the others didn't, some of the time.
• Some customers changed lines after I started timing them. I counted them under "None of these," which may be why that line is so slow. Or maybe people who switch lines really are slower. More research is needed.

[Mom's note: My own method is to pick the cashier who has been there the longest. Nathan says this is "not testable." I say it's worked for twenty years.]

Experiments I Was Not Allowed to Do

Coming up: Experiments Planned for Winter Break including Toast, Part II (in the garage) and a Y2K test of every clock in the house

Observers recorded:

n = 3,527 since August 1999

Questions, data, peer review:
[email protected]

Last updated Tuesday, September 21, 1999