Marble Collision Experiment: Where Does the Energy Go?

Line up five marbles so they touch. Roll a sixth into the end. One marble leaves the other side and the middle four do not move. Roll two in, and two come out. Every time.

Momentum alone cannot explain that, which is what makes this collision experiment with marbles worth more than thirty seconds of amusement. And the click you hear is not a sound effect. It is the energy that failed to make it through, and you can measure how much.

Key Takeaways

• You will roll marbles into a touching line, then measure the coefficient of restitution by bouncing one on the floor.
• The concept is that a collision has to satisfy two bookkeeping rules at once: momentum and energy.
• Best for ages 8 to 14. Rolling marbles suits any age; the two-rule argument lands around 11.
• The practical detail: identical marbles on a level, hard, straight channel, or the pattern falls apart.
• Marbles are a severe choking hazard for children under three. Count them out and count them back in.

What You’ll Need

Item	How much	Where to get it	If you don’t have it
Glass marbles, identical	8	Toy shop, sold loose and cheap	Identical ball bearings, or dried chickpeas for a much duller version
Two rulers or straight battens	2	Pencil case or hardware shop	A length of curtain track, or a folded card channel
Masking tape	1 roll	Drawer	Blu-tack at both ends of each ruler
A hard, level table	1	Kitchen	A tiled floor, which is often flatter than a table
Tape measure	1	Toolbox	Marked string
Hard floor for bounce tests	1	Tiled kitchen or hallway	A thick wooden board laid on the floor
Phone with slow-motion video	1	Pocket	You can do the whole thing by eye; the video makes the bounce height easier
One odd larger marble	1	Toy shop	Any ball noticeably heavier than the others
Notebook	1	Anywhere	Five bounces per surface is a lot of numbers

Time: prep 10 min, experiment 40 min, cleanup 10 min, which includes finding every marble Cost: a bag of marbles Mess level: low, but marbles on a floor are a slip risk

Do the Experiment

Step 1 — Build a straight, level channel

Tape two rulers to the table, parallel, with a gap slightly narrower than a marble’s diameter so the marbles sit in the groove. Check the table is level by placing one marble in the middle and letting go; if it rolls, shim the table or move to the floor.

Step 2 — Line up five marbles touching

Push five marbles together in the middle of the channel so each one is in contact with its neighbours. Contact matters. A 1 mm gap turns one collision into a sequence of separate ones and blurs the result.

Step 3 — Roll one in and watch the far end

Roll a single marble in from the left, gently and along the groove. Count how many leave the right-hand end and watch the middle ones. Write down what you saw before you do anything else, because the surprise fades fast.

Step 4 — Roll two in, then three

Hold two marbles together and release them as a pair. Then three. Each time, count the number leaving. Then try rolling one in fast and one in slow, and note whether the number leaving changes or only the speed does.

Step 5 — Measure how much energy survives

Drop a single marble from exactly 50 cm onto your hard floor and measure the height of the first bounce. Five drops, median taken. Divide the bounce height by the drop height and take the square root; that number is the coefficient of restitution for marble on floor.

Step 6 — Measure the loss through the chain

Roll a marble along the empty channel from a fixed starting point and mark where it stops. Then put the five-marble line back and roll from the same point; mark where the exiting marble stops. The difference is energy that went into the collisions rather than into motion.

What you should see: one in gives one out, two give two, three give three, and the middle marbles stay put. The exiting marble is clearly slower than the incoming one, and you can hear why: a sharp click at the moment of impact. On the bounce test, a glass marble dropped 50 cm onto ceramic tile typically rebounds 35 to 42 cm, giving a restitution coefficient of roughly 0.84 to 0.92. On wood it will be lower; on carpet it barely bounces at all.

What’s Actually Going On in a Collision Experiment With Marbles

Two separate accounting rules apply to every collision, and the interesting part is that you need both.

The first is momentum. In an isolated system, where no outside horizontal forces interfere, the total momentum before a collision equals the total momentum afterwards (The Physics Classroom on momentum conservation). Momentum is mass times velocity, added up with direction taken into account (The Physics Classroom on isolated systems).

Here is why that rule alone is not enough. Suppose one marble arrives at 1 m/s. Momentum would be perfectly satisfied if two marbles left at 0.5 m/s each. It would also be satisfied if four left at 0.25 m/s each. Momentum does not pick between these. So why does exactly one come out?

Because of the second rule. In a nearly elastic collision, kinetic energy is also nearly conserved, and kinetic energy depends on the square of speed. One marble at 1 m/s carries a certain amount. Two marbles at 0.5 m/s each carry only half as much, because halving the speed quarters each marble’s energy and you only have twice as many marbles. So the two-out solution would conserve momentum while destroying half the energy, which near-elastic glass does not do. The only outcome that satisfies both rules is one marble out at nearly the incoming speed (HyperPhysics on elastic and inelastic collisions).

That is the real content of the demonstration, and why “one in, one out” feels magical until you see that two constraints are doing the work.

Now the honest part, which HyperPhysics states plainly: no large-scale collision is perfectly elastic. Any macroscopic impact converts some kinetic energy into internal energy and other forms, and collisions between hard spheres are merely nearly elastic. Your marbles obey the rule approximately, which is why the exiting marble is slower than the incoming one.

Where did the missing energy go? Mostly into sound and heat inside the glass, plus a little vibration in the marbles and table. The click is literally the escaping energy, converted into a pressure wave in the air. This is the sentence worth saying out loud: the noise is the loss. Quieter collisions are more elastic, which is why a dropped steel ball bearing rings and rebounds high while a dropped lump of putty thuds and stays put.

The coefficient of restitution puts a number on all of it. It is the ratio of separation speed to approach speed, and for a ball bouncing off a fixed surface it equals the square root of bounce height over drop height. Rod Cross has published a series of papers on measuring it in Physics Education, including the awkward case of a ball striking at an angle (Cross, 2020). Square your measured coefficient and you have the fraction of energy that survived one impact. A coefficient of 0.9 means 81 percent survives, so 19 percent is lost per bounce.

One subtlety your rolling marbles introduce, and it is worth naming rather than hiding. A rolling marble stores part of its kinetic energy in spin, not just in forward motion, and a collision between smooth spheres transfers forward motion much more readily than spin. That is one reason the chain is less tidy than a hanging Newton’s cradle, where the balls swing instead of rolling. The bounce-height method avoids the issue entirely, which is why it is the measurement to trust.

How to Explain It to Your Kid

Ages 5–8: The push goes through and comes out the end

Have them stand in a row with two friends or family members, each with hands on the shoulders of the next, and gently push the one at the back. The push travels along and the one at the front steps forward. Then let them roll the marble. Say the push goes through all the middle marbles and comes out of the last one, and that the click is the sound of a tiny bit of the push getting lost.

Ages 9–12: Two rules at the same time

Give them the puzzle before the answer. If one marble goes in, why not two out at half speed? Let them see that the amount of pushing works out either way. Then tell them the second rule: the collision also has to keep almost the same amount of movement energy, and energy counts speed twice over. Two marbles at half speed have only half the energy, so that answer is not allowed.

Ages 13+: Solve both equations and measure the coefficient

Write conservation of momentum and conservation of kinetic energy for a one-dimensional collision between equal masses and solve them together. The result is that the incoming ball stops and the struck ball leaves with the incoming velocity. Then measure the restitution coefficient from bounce heights on three surfaces, square it to get energy retention, and compare the chain’s energy loss with the per-collision loss multiplied over the chain length.

The question to ask: “If we glued all five marbles together into one solid rod, what would come out when you roll one in?” (Nothing comes out, because there is no last marble to leave. The whole glued rod would move slowly instead, conserving momentum but losing a lot of energy. It tests whether they understand that the separate bodies are what make the elastic chain work.)

Why Teaching This Early Pays Off

The transferable idea here is that a physical situation can be pinned down by more than one conservation law at once, and that the answer is whatever satisfies all of them. That pattern recurs across physics and chemistry: conservation of mass and charge in a reaction equation, energy and momentum in particle physics, mass and energy in a flow. A child who has watched two rules jointly rule out three plausible outcomes has met the pattern in a form they can hold.

The second payoff is the noise. Most children assume energy “runs out.” Hearing that the click is the lost energy is a cleaner introduction to dissipation than any diagram, and it makes the whole category of inefficiency, from warm phone chargers to squealing brakes, feel like one thing.

As for how to run it: do not just hand over the marbles and leave. Alfieri and colleagues’ meta-analysis of discovery learning found that unassisted discovery did not help learners, while discovery supported by explanation, feedback or worked examples did (Alfieri et al., 2011). The useful moment is asking why not two out at half speed, and then supplying the energy rule when they get stuck.

No single afternoon makes a physicist and it is better not to pretend. If this one works, our guides to dominoes amplifying energy down a chain and the egg drop as an impulse measurement build on the same bookkeeping.

Before You Start: What Parents Should Know

Best age	8 to 14. Younger children can roll marbles with an adult present, but not unsupervised
Supervision	Constant if there is a toddler, a crawling baby or a dog in the house. Otherwise light
Real hazards	Choking. A marble is almost exactly the worst possible size for a small child’s airway, and dogs swallow them too. Also slipping on a marble underfoot, and chipped glass from a dropped marble on tile
Don’t	Don’t do this on a floor where a marble can roll under furniture and be forgotten, and don’t use chipped marbles. A chipped glass marble sheds small sharp fragments

Count the marbles into a cup at the start, write the number down, and count them back in at the end. If one is missing, find it before anyone else comes into the room. That is not fussiness: small round objects are among the most common causes of serious choking in young children, and the research on it is unambiguous.

When It Doesn’t Work

Marbles jump out of the channel. The gap between your rulers is too wide, so the marbles ride up. Narrow the gap until each marble sits down into the groove with its equator above the rulers. Tape both rulers down at several points so they cannot spread.

The whole line drifts to one side. Your table is not level. Test with one marble left alone in the middle; if it moves, you are measuring gravity instead of collisions. Shim the table legs or move the whole experiment to a tiled floor.

Two marbles come out when you rolled one in. Check that all the marbles are identical and that they were truly touching. Mismatched masses change the solution to the two conservation equations, and gaps turn one collision into several. Both effects muddy the pattern.

The exiting marble barely moves. Either the channel has friction, which long rulers with a rough edge certainly do, or you rolled too gently. Roll harder and shorten the channel. If the marble still crawls, your surface is absorbing the energy; try a tile floor.

The bounce test gives wildly different heights. You are judging the peak by eye, which is genuinely hard. Film it at high frame rate against a ruler taped to a wall and read the top frame, or drop against a wall with marks every 2 cm and have a second person watch at eye level.

Take It Further

Build a restitution table. Measure the coefficient on ceramic tile, wood, a hardback book and a sheet of paper over tile, five drops each. Square each value to get the fraction of energy retained. You will find the paper barely changes the tile result, which is itself a good discussion about how much of the loss happens inside the marble.

Vary the chain length. Measure the exiting marble’s travel distance for lines of 1, 3, 5 and 7 marbles, rolling from a fixed start each time. Energy is lost at each contact, so the exit distance should fall as the chain gets longer. Predict the shape of the fall before measuring.

Break the symmetry on purpose. Roll your one large, heavy marble into a line of small ones, then reverse it and roll a small one into a line that ends with the large one. The outcomes are different and both follow from solving the same two conservation rules with unequal masses. Working out which one bounces back is a genuinely good piece of reasoning.

What to Watch For Over the Next 3 Months

• Week 4: Your kid describes a collision in terms of what is conserved, rather than which object is stronger.
• Month 2: The word “energy” appears attached to sound or heat rather than only to motion. Listening for how loud something is when it lands becomes a habit.
• Month 3: They ask about a Newton’s cradle, a pool table or a snooker break. All three are the same two rules, and a pool table is the best natural laboratory for them that exists.

Frequently Asked Questions

Why does only one marble come out?

Because the collision has to conserve both momentum and, very nearly, kinetic energy. Two marbles leaving at half speed would conserve momentum but destroy half the energy, which near-elastic glass does not do. One marble leaving at nearly the incoming speed is the only outcome that satisfies both.

What makes the clicking sound?

Energy that did not survive the collision, converted into a pressure wave in the air. Some also becomes heat and vibration inside the glass. Quieter collisions are more elastic, which is why a steel ball rings and rebounds high while putty thuds and stops.

How do we measure the coefficient of restitution?

Drop a marble from a measured height onto a hard surface and measure the first bounce height. The coefficient is the square root of bounce height divided by drop height. Square the coefficient and you get the fraction of energy that survived the impact.

Does it matter that the marbles are rolling?

Yes, a little. A rolling marble keeps some energy in spin, and smooth spheres transfer forward motion much more readily than spin, which makes the rolling chain less tidy than a hanging Newton’s cradle. The bounce-height measurement sidesteps the issue, which is why it is the number to trust.

What age is safe for this?

Eight and up with an adult nearby, and not at all in a house with a crawling baby, a toddler or a dog unless every marble is accounted for continuously. A marble is close to the worst possible size for a small airway, and that is the single real hazard here.

---

About the author

Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.

---

Sources

1. Nave, R. “Elastic and Inelastic Collisions.” HyperPhysics, Georgia State University. http://hyperphysics.phy-astr.gsu.edu/hbase/elacol.html
2. Nave, R. “Inelastic Collisions.” HyperPhysics, Georgia State University. http://hyperphysics.phy-astr.gsu.edu/hbase/inecol.html
3. The Physics Classroom. “Momentum Conservation Principle.” https://www.physicsclassroom.com/class/momentum/Lesson-2/Momentum-Conservation-Principle
4. The Physics Classroom. “Isolated Systems.” https://www.physicsclassroom.com/class/momentum/Lesson-2/Isolated-Systems
5. Cross, R. (2020). “Coefficient of restitution for an obliquely bouncing ball.” Physics Education, 56(1), 015004. https://doi.org/10.1088/1361-6552/abbcbd
6. Alfieri, L., Brooks, P. J., Aldrich, N. J., & Tenenbaum, H. R. (2011). “Does discovery-based instruction enhance learning?” Journal of Educational Psychology, 103(1), 1–18. https://doi.org/10.1037/a0021017