Drop a Thousand Marbles: lesson plan
Students pour a thousand marbles through a Galton board, predict where single marbles will land, and use coin flips and Pascal’s triangle to explain why random bounces always pile up into the same bell-shaped curve.
- Ages
- 11–16 (grades 6–10)
- Time
- 50 minutes
- Subject
- Maths: probability, the binomial distribution and the normal curve
- You need
- Devices with a browser (one per pair), coins (12 per pair or one to flip 12 times), squared paper, calculators, Pascal’s triangle worksheet (optional)
Open Drop a Thousand Marbles Jump to the quiz
Learning objectives
By the end of the lesson, students can:
- Explain why a single marble’s path is unpredictable but the shape made by many marbles is predictable.
- Use Pascal’s triangle to count the routes to each bin of a 12-row board and calculate the probability of landing in a given bin.
- Compare experimental results with theoretical probability and explain why they get closer as the number of trials grows.
- Describe the normal distribution and give real examples of quantities that roughly follow it.
Key vocabulary
- Probability
- How likely an event is, from 0 (impossible) to 1 (certain). A fair coin landing heads has probability ½.
- Outcome
- One possible result of an experiment, such as a marble landing in bin 4.
- Pascal’s triangle
- A triangle of numbers where each is the sum of the two above it. Row 12 (1, 12, 66, 220, 495, 792, 924, …) counts the routes to each bin.
- Binomial distribution
- The probabilities of getting 0, 1, 2 … successes in a fixed number of independent yes/no trials, such as left or right bounces.
- Normal distribution
- The smooth, symmetrical bell-shaped curve that many real measurements follow.
- Relative frequency
- How often something happened divided by the number of trials. It estimates the probability.
- Central limit theorem
- The result that adding many small, independent random effects tends to give a normal distribution.
Background for the teacher
The board in the game has 12 rows of pegs and 13 bins. At each peg a marble goes roughly left or right with equal chance, so its final bin depends only on how many of its 12 bounces went right. There are 2¹² = 4,096 equally likely left–right routes. The number that end in each bin is a row of Pascal’s triangle: 1, 12, 66, 220, 495, 792, 924, 792, 495, 220, 66, 12, 1. So the centre bin has probability 924/4,096 ≈ 22.6%, each edge bin 1/4,096 ≈ 0.024%, and the three middle bins together about 61%. Out of 1,000 marbles you expect about 226 in the centre, about 193 in each neighbour and about 121 in the next, but almost never one at an edge.
The mean bin is the centre and the standard deviation is √(12 × ½ × ½) ≈ 1.7 bins. This is a binomial distribution; with more rows it approaches the normal curve, a consequence of the central limit theorem. The game simulates real marbles with gravity, bouncy pegs (coefficient of restitution about 0.3) and collisions, so bounces are not perfect coin flips, just like a real board. The red dotted line shows the exact prediction. The tilt slider (−20° to +20°) biases every bounce, which shifts and skews the curve. Francis Galton described the board, which he called the quincunx, in his 1889 book Natural Inheritance.
Warm-up (5 minutes)
- Ask: "If you flip a coin 12 times, how many heads do you expect? Would you be surprised by 12 heads? By 6?"
- Show a picture of a Galton board or a Plinko board. Ask: "Where do you think most balls end up? Why?" Students sketch their predicted pile.
- Quick vote: is it possible to predict where a single marble will land? Is it possible to predict the shape that 1,000 marbles make?
Activity with the game
- Pairs open Drop a Thousand Marbles. Before pouring, they mark on squared paper how many of 1,000 marbles they expect in each of the 13 bins.
- Press Pour 1,000 and watch. Pairs compare the pile with their sketch and with the red dotted prediction line. Which bins did they over- or underestimate?
- Coin model: each pair flips a coin 12 times and counts heads (heads = right bounce). Repeat 10 times and record the head counts. Pool the class results into a tally chart for 0 to 12 heads and compare with the game’s pile.
- Count the routes: on a 4-row board first (16 routes, bins 1, 4, 6, 4, 1), pairs trace every route by hand. Then build Pascal’s triangle to row 12 and calculate the probability for the centre bin (924/4,096) and an edge bin (1/4,096).
- Call a bin: each pair uses Call a bin ten times, choosing a bin and dropping one marble. Record hits. Discuss why even the "best" bin is wrong most of the time (about 77% of the time for the centre).
- Tilt test: set the tilt slider to +10° and pour again. Describe how the curve moves and whether it stays symmetrical. Try −10°.
Discussion questions
- Why do most marbles land in the middle bins?
Show a model answer
Many more routes lead to the middle. To reach the centre a marble needs six rights and six lefts in any order, and there are 924 ways to arrange that. To reach an edge it needs twelve bounces the same way, and there is only one route.- If the centre bin is the most likely, why did calling it fail most of the time?
Show a model answer
Its probability is only about 22.6%, so about 77% of single marbles land somewhere else. "Most likely" does not mean "likely".- Why did the pile match the red line better after 1,000 marbles than after 50?
Show a model answer
With few trials, chance makes the relative frequencies jump about. As the number of trials grows, relative frequency settles towards the true probability (the law of large numbers), so the shape smooths out.- What happened when you tilted the board?
Show a model answer
Every bounce became more likely to go one way, so the probability of a right bounce was no longer ½. The whole curve slid towards the low side and became lopsided (skewed), though it still had one peak.- Where do bell curves appear in real life?
Show a model answer
Adult heights, measurement errors, the weights of apples from one tree and many test scores are roughly normal, because each is the sum of many small, independent effects, like bounces on the pegs.
Common misconceptions
- If a marble went right six times in a row, it is "due" to go left.
- Each bounce is independent; the pegs have no memory. This mistake is called the gambler’s fallacy.
- Every bin is equally likely, because each bounce is 50/50.
- Each bounce is 50/50, but the bins are reached by very different numbers of routes. The centre bin has 924 routes; an edge bin has one.
- An experiment that does not match the prediction exactly must be wrong.
- Random results always scatter around the prediction. The match improves with more trials, but is never perfect.
Quick quiz
How many different left–right routes are there through 12 rows of pegs?
- 24
- 144
- 4,096
- 1,000,000
Answer
C. Each row doubles the routes: 2¹² = 4,096.
How many routes lead to the centre bin of a 12-row board?
- 12
- 66
- 494
- 924
Answer
D. The middle number of row 12 of Pascal’s triangle is 924 (12 choose 6).
What is the probability that a marble lands in one particular edge bin?
- 1/13
- 1/24
- 1/144
- 1/4,096
Answer
D. Only one route out of 4,096 goes right (or left) every time.
You pour 1,000 marbles. About how many land in the centre bin?
- About 77
- About 226
- About 500
- About 924
Answer
B. 1,000 × 924/4,096 ≈ 226.
In Pascal’s triangle, how is each number made?
- By doubling the number above
- By adding the two numbers above it
- By multiplying the row number by 2
- By counting the bins
Answer
B. Each entry is the sum of the two entries diagonally above. Row 3 is 1, 3, 3, 1, so row 4 is 1, 4, 6, 4, 1.
What happens to the curve when you tilt the board to the right?
- It disappears
- It shifts right and becomes lopsided
- It becomes flat
- Nothing changes
Answer
B. A right bounce becomes more likely than ½, so the whole distribution shifts and skews.
Going further
- Calculate: use the Galton board probability calculator to find the chance of landing in each bin for boards with 6, 12 and 20 rows. How does the centre bin’s probability change?
- Real data: measure the hand span of everyone in the class (or a larger group) and draw a frequency chart. Is it roughly bell-shaped? Where is the peak?
- Build a mini Galton board from a cork board, drawing pins and rice or lentils, and compare its pile with the game.
Calculators for this lesson
Curriculum links
- England National Curriculum, KS3 Maths (Probability): record, describe and analyse the frequency of outcomes of probability experiments; enumerate sets and use systematic listing; understand that the probabilities of all possible outcomes sum to 1.
- Common Core 7.SP.C.6: approximate the probability of a chance event by collecting data on the chance process that produces it and observing its long-run relative frequency.
- Common Core 7.SP.C.7: develop a probability model and use it to find probabilities of events; compare probabilities from a model to observed frequencies.
- GCSE Maths (Probability): relative frequency and expected outcomes; the effect of the number of trials on how closely results match theory.
The science in this plan matches the notes and sources under Drop a Thousand Marbles. Free to use and adapt for teaching. Ideas or corrections: hello@notebookfun.com.