Spaghetti Marshmallow Tower Experiment: Why It Falls

Here is the number that settles most arguments about tower building: a spaghetti strut half as long carries roughly four times the load before it buckles. Not twice. Four times, because buckling load falls with the square of length. Every tall, elegant, long-legged spaghetti tower your child builds is fighting that relationship and losing. This spaghetti marshmallow tower experiment is usually run as a race to be tallest; run it with a tape measure, a kitchen scale and that one piece of physics, and it becomes a genuine engineering investigation.

Key Takeaways

• Build towers from dry spaghetti and marshmallows, then measure height, mass and the load each one carries before collapsing.
• It teaches compression and tension, and specifically that slender members in compression fail by buckling, not by crushing.
• Fits ages 8 to 15. The measuring and re-building is what makes it work for older kids.
• The usual failure: one tower, built slowly, tested once. Three fast towers beat one careful one, every time.
• Wheat allergen, gelatin in marshmallows, and snapped spaghetti ends are genuinely sharp.

What You’ll Need

Item	How much	Where to get it	If you don’t have it
Dry spaghetti	20 strands per attempt, 60+ total	Supermarket	Dry linguine or bucatini. Contains wheat
Mini marshmallows	20 per attempt	Supermarket	Modelling clay or plasticine is better for load tests. Gumdrops work. Marshmallows contain gelatin and are a choking hazard under 3
Tape measure or metre rule	1	Toolbox	A 30 cm ruler used carefully, twice
Kitchen scale	1	Kitchen	Count coins into a cup instead
Small paper cup	1	Kitchen	A bottle cap for small loads
Coins or washers	50+	Change jar	Dried beans, counted
Timer	1	Phone	Any clock
Paper and pencil	1 sheet	School supplies	Record every attempt, including the bad ones

Time: prep 5 min, build and test 45 min, cleanup 10 min Cost: a packet of spaghetti and a bag of marshmallows Mess level: medium. Snapped spaghetti goes everywhere and hurts bare feet

Do the Experiment: The Spaghetti Marshmallow Tower Experiment

Step 1 — Run the single-strut test first, before any tower

Hold one full-length strand of spaghetti vertically on the table and press straight down on top with one finger. It bows sideways and snaps almost immediately. Now snap a strand into four pieces and press down on one short piece. It takes a lot more force. Ask your child to rank, by feel, how much more. Then tell them: roughly four times for half the length.

Step 2 — Take a prediction about towers

Ask: “Will the tallest tower be the one that holds the most weight?” Write down the answer, and also ask for a guess at the tallest height achievable with 20 strands. Most children predict 60 cm or more. Realistic answers are 30 to 50 cm for a tower that still stands when you let go.

Step 3 — Build three towers in 15 minutes, not one in 45

Set a timer for five minutes per tower and build three. Rule: the tower must stand on its own and hold one marshmallow on top. Fast, scrappy, and tested immediately. This instruction matters more than any building tip, and the reason is in the research below.

Step 4 — Measure everything about each tower

For each one, record three numbers: height in centimetres, total mass of the structure in grams, and the load it carried before collapse. To measure load, balance the paper cup on top and add coins one at a time until it fails. Weigh the cup and coins to get the failure load.

Step 5 — Build a fourth tower with short braced triangles

Now build deliberately: break the spaghetti into shorter pieces, roughly 7 to 10 cm, and assemble a triangulated frame, a triangular base, diagonal braces on every face, no long unsupported members. It will look stubby and inelegant. Measure the same three numbers.

Step 6 — Compare, and compute load per gram

Build the table. Then compute load divided by structure mass for each tower. That ratio is the one engineers care about, and it will almost always favour the short braced tower, even though it is the shortest and least impressive-looking thing on the table.

What you should see: towers of 30 to 50 cm from 20 strands, carrying anywhere from 20 to 200 grams depending entirely on bracing. Long-legged towers fail suddenly and sideways, with one leg bowing outward just before it goes. Marshmallow joints also creep — they slowly deform under sustained load, so a tower that stands at minute one can sag by minute ten. That’s not your child’s fault, and it is a real material behaviour worth naming.

What’s Actually Going On

Every member in a structure is doing one of two jobs. In tension, it is being pulled, and it resists by the material’s strength. In compression, it is being pushed, and here something counter-intuitive happens: a long slender member does not fail by being crushed. It fails by suddenly bending sideways, an instability called buckling.

Leonhard Euler worked out the governing relationship in the eighteenth century. The critical buckling load of a slender column is proportional to EI/L², where E is the material’s stiffness, I is the second moment of area of the cross-section, and L is the length. The length is squared and in the denominator, which is the whole game. Halve the length, quadruple the load capacity. Use a strand half as thick instead and you lose far more, because I for a circular section goes with the fourth power of diameter.

So the single most effective thing your child can do to a spaghetti tower is make the compression members shorter. Not stronger, not thicker, not more numerous. Shorter. Bracing achieves exactly this: a diagonal brace halfway along a leg converts one long column into two short ones, quadrupling what that leg can carry, for the cost of one extra piece of spaghetti.

The second idea is triangulation. A marshmallow joint can rotate, which makes it a pin joint in engineering terms. A rectangle of four pinned members can collapse into a parallelogram without any member changing length at all — it has no resistance to being pushed sideways. A triangle cannot: changing its shape requires changing a side length. That’s why every crane, pylon, roof truss and bridge frame you have ever seen is made of triangles, and why a square-based spaghetti tower racks over while a triangulated one doesn’t.

Two more honest notes. First, spaghetti’s fracture behaviour is itself a published physics result. Audoly and Neukirch explained in Physical Review Letters why a dry spaghetti strand bent until it breaks almost never breaks into exactly two pieces: the first break sends a flexural wave along the fragments, which locally increases curvature and triggers further breaks — a cascade. Heisser and colleagues later showed in PNAS that twisting a strand before bending it suppresses the cascade and lets you break it in two. Your child can test both claims on the offcuts.

Second, the marshmallows. They are viscoelastic, which means they deform progressively under a constant load rather than reaching a fixed shape. For a speed challenge they are perfect. For measuring failure loads they introduce a time-dependent wobble into every reading. If your child wants a clean load test, switch the joints to small balls of modelling clay and say why you did.

How to Explain It to Your Kid

Ages 5–8: Short legs are strong legs, and triangles don’t wobble

Give them one long piece of spaghetti to push on, then a short piece. Let them feel the difference with their own hands before you say a word. Then make a square from four pieces and four marshmallows and push it sideways — it flops. Add one diagonal piece and push again. It holds.

That’s the whole lesson at this age, and it’s a big one: short legs and triangles. Let them build whatever they like afterwards, and when it falls, ask which leg was longest.

Ages 9–12: Buckling, and the squared rule

Introduce buckling by name and demonstrate it: a strand that bows sideways rather than crushing. Then give the rule in words (halve the length, four times the load) and test it properly. Cut strands to 20, 10 and 5 cm. For each length, stand it on a kitchen scale and press down carefully until it buckles, noting the maximum reading.

Three lengths, three loads. If the 10 cm strand took about four times what the 20 cm one did, the rule is holding. It won’t be exact, because end conditions and tiny bends matter, and explaining why real data is messier than a formula is more valuable than a clean result.

Ages 13+: Design iteration and the thing most teams get wrong

With a teenager, run this as a design-process study. Give them three five-minute rounds and require them to measure and change one thing between rounds. Then compare against a single twenty-minute build. The three-round version almost always wins on load per gram.

There’s a reason, and it has a name. Jansson and Smith’s work on design fixation in Design Studies showed that designers who commit early to one solution conform to it even when it is poor, and that exposure to a single example narrows the solutions people generate. Teams that spend all their time planning one tower have fixated before they have any data. Teams that build, break, measure and rebuild are doing what the National Research Council’s framework calls optimising a design solution.

The question to ask: “If I gave you ten more pieces of spaghetti but told you the tower had to be the same height, where would you put them, and how much more load should it hold?”

Why Teaching This Early Pays Off

The buckling idea is one of those rare facts that is simple, quantitative, and immediately visible in the world: scaffolding has cross-braces, pylons are triangulated, bike frames are triangles, tall thin things get fat near the bottom. A child who owns the 1/L² relationship has a tool they will use every time they build anything, for the rest of their lives.

It also teaches iteration properly, which is harder than it sounds and is the actual working method of engineering. Most school design tasks are run as one build and one test, which teaches planning but not learning-from-failure. Three timed rounds with measurements between them teaches the opposite, and the measurement is what separates it from just playing. Our pieces on the paper bridge experiment and building a catapult extend the same habit to other structures.

As always, the modest claim is the true one: an afternoon of spaghetti doesn’t change academic trajectories and nothing in the literature says it does. What it leaves is a quantitative rule, a demonstrated preference for iteration over planning, and a child who, next time someone builds something tall and thin, will look at the longest member first.

Before You Start: What Parents Should Know

Best age	8 to 15. Younger children enjoy the building but can’t yet hold the measure-and-rebuild discipline that makes it an experiment.
Supervision	Light, but an adult runs the timer and the load test and clears up snapped spaghetti promptly.
Real hazards	Snapped dry spaghetti has sharp points and fragments fly when strands break. Wheat allergen. Marshmallows are a choking hazard for under-threes and contain gelatin.
Don’t	Don’t put faces close to a tower during the load test. Don’t let anyone eat the marshmallows that have been handled and loaded. Don’t test on a hard floor in bare feet.

Two practical notes. When spaghetti snaps under load it does so with a crack and the fragments can travel a metre or more, so everyone should be at arm’s length and nobody should be leaning over the top. And clear the floor between rounds: dry spaghetti shards in a carpet are a genuine foot hazard, and they turn up for weeks. A tray or a large sheet of card under the build area solves both problems.

When It Doesn’t Work

• Every tower collapses as soon as you let go. The base is too small. A tower needs a base roughly a third of its height to stay up without support, and it needs diagonals on the bottom section.
• The marshmallows slide off the spaghetti. They’re too dry or too fresh. Slightly stale marshmallows grip best. Or switch to modelling clay, which grips reliably and doesn’t creep.
• It stands at first and sags over ten minutes. Marshmallow creep, which is a real material property, not a build error. Test within two minutes of finishing, or use clay joints.
• The spaghetti shatters while building. It’s brittle and it shatters by cascade, as the physics papers describe. Snap it deliberately against a table edge rather than bending it slowly, and accept the breakage as part of the material budget.
• Your child builds the same tower three times. Design fixation, right on cue. Require one specific change per round: shorter members, then an added diagonal, then a wider base. Naming the change is the whole intervention.

Take It Further

Measure the buckling law directly. Cut strands to 20, 15, 10 and 5 cm. Stand each vertically on a kitchen scale with a flat object on top and press down slowly until it buckles, recording the peak reading. Plot failure load against length, then against 1/length². If the second plot is closer to a straight line than the first, you have experimentally recovered a result from 1757 with a pasta packet.

Run a fair competition with a real metric. Instead of “tallest”, score towers on load carried divided by mass of structure, measured identically. Give three builders the same 20 strands and 20 joints. The winner will not be the tallest, and the argument afterwards about whether height should count at all is exactly the argument engineers have about design requirements.

What to Watch For Over the Next 3 Months

• Week 4: They add a diagonal to something unprompted — a den, a cardboard box, a shelf. That’s triangulation transferring.
• Month 2: They start noticing cross-bracing in the built world: scaffolding, pylons, gates, the back of a flat-pack bookcase.
• Month 3: They reach for a second attempt faster than before, and talk about what they’ll change rather than restating the plan. That shift is the most valuable outcome here.

Frequently Asked Questions

Why do tall spaghetti towers always fall over?

Two reasons working together. Long compression members buckle at loads proportional to one over length squared, so tall legs are drastically weaker. And marshmallow joints rotate, so an untriangulated frame can deform into a parallelogram without any member breaking at all. Shorter members and diagonal braces fix both.

Are marshmallows or modelling clay better?

Marshmallows for a timed build challenge, clay for measuring loads. Marshmallows are viscoelastic and creep under sustained load, so a tower sags over minutes and your failure-load readings drift. Clay grips better and holds its shape, which makes the comparison between towers fair.

Why does dry spaghetti never snap into exactly two pieces?

Because the first break sends a flexural wave along the fragments that locally increases curvature and triggers further breaks — a cascade, explained in a 2005 Physical Review Letters paper. A later PNAS study showed that twisting a strand before bending suppresses the cascade and allows a clean break in two.

Should we aim for tallest or strongest?

Pick one and say so before you start, because they conflict. Tallest rewards long members, which buckle. Strongest rewards short braced members, which are stubby. Scoring on load carried per gram of structure is the most informative version and the closest to how real design requirements work.

---

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. Audoly, B., & Neukirch, S. (2005). “Fragmentation of Rods by Cascading Cracks: Why Spaghetti Does Not Break in Half.” Physical Review Letters, 95, 095505. https://doi.org/10.1103/PhysRevLett.95.095505
2. Heisser, R. H., Patil, V. P., Stoop, N., Villermaux, E., & Dunkel, J. (2018). “Controlling fracture cascades through twisting and quenching.” PNAS, 115, 8665–8670. https://doi.org/10.1073/pnas.1802831115
3. Jansson, D. G., & Smith, S. M. (1991). “Design fixation.” Design Studies, 12, 3–11. https://doi.org/10.1016/0142-694X(91)90003-F
4. Gordon, J. E. (1978). Structures: Or Why Things Don’t Fall Down. Penguin.
5. Massachusetts Institute of Technology OpenCourseWare. “3.11 Mechanics of Materials.” https://ocw.mit.edu/courses/3-11-mechanics-of-materials-fall-1999/
6. National Research Council. (2012). A Framework for K-12 Science Education. National Academies Press. https://doi.org/10.17226/13165