Reversing a garden sprinkler is now a serious fluid-dynamics puzzle

The garden toy is a familiar piece of summer clutter: a plastic cross with two or three curved arms, each pierced by a nozzle that spits a thin arc of water into the air. Spin one around and it whirs back the other way. The motion, and the shape of the water jet that produces it, has now been turned inside out by a small group of mathematicians who report that the toy version of the device still hides a 40-year-old puzzle in fluid dynamics.

The work, led by a team at University College London and New York University, reverses the usual orientation of the so-called "silly sprinkler." Instead of pumping water through a spinning head and watching the recoil spin it faster, the researchers clamp the head still and drive the jets past it, using the geometry of the spray to extract information that the original forward-running experiment has never quite yielded. The result, they argue, is a cleaner window onto an old question: how the angular velocity of the arms scales with flow rate as the pressure drops.

A toy with a long paper trail

The reverse-sprinkler problem has its own entry in the history of physics. A now-famous thought experiment by Richard Feynman, often recounted in lectures and in a 1985 memoir, asked which way a sprinkler would spin if it were immersed in a fluid and the fluid were made to flow past it rather than pumped through it. The intuition has tripped up generations of undergraduates: a naive application of momentum conservation seems to predict rotation in the opposite direction to the forward case, but the answer, the careful argument goes, is that the sprinkler does not rotate at all.

That Feynman's puzzle and the simpler "silly sprinkler" on the lawn are not the same problem is a distinction the new paper draws explicitly. The garden toy operates in air, with discrete jets. The thought experiment involves a body fully bathed in moving fluid. But the two share enough geometry, the authors write, that the toy has become a useful surrogate for studying low-flow-rate behaviour, where the jets of a real lawn sprinkler stop behaving like the idealised streams that textbook calculations assume.

What the reverse rig reveals

The laboratory rig inverts the device. Instead of squirting water out of nozzles fixed to a rotor, the team feeds dyed water into stationary nozzles aimed so the emerging jet grazes a stationary arm, then maps the path of the jet downstream. By varying the flow rate and the angle of the nozzles relative to the arms, they can probe the regime in which the jets begin to break up, curve, and merge with their neighbours. In the forward case that regime is hard to access because the spin of the head itself perturbs the very jets the experimenter is trying to measure.

Two findings stand out in the early data. First, the jets leaving a real nozzle do not follow the straight lines assumed in the textbook derivation; they curve noticeably under gravity and under the influence of surrounding air. Second, the transition between smooth jet flow and a chaotic, droplet-filled regime occurs at a flow rate that matches what a handful of older, less precise measurements had suggested but that no forward-running experiment had ever cleanly captured. The researchers describe the agreement as "suggestive rather than conclusive" and flag the low-flow regime as the place where the discrepancy between theory and experiment is widest.

A field where the small effects matter

Sprinkler geometry is not only a curiosity. The same balance of forces shows up in any device that sprays a fluid through a curved channel: turbine blade cooling, agricultural irrigation, the tiny nozzles that fire droplets in industrial printing. In each case, the assumption that the jet leaves the nozzle exactly along its axis is what lets engineers simplify the flow. Where the assumption breaks down, at low pressure or in devices with sharply curved arms, predictions about torque and flow distribution start to drift.

The new work belongs to a slow tradition of using the toy sprinkler as a teaching tool that occasionally sharpens into a research instrument. Earlier this century, separate groups at the University of Minnesota and at Princeton reported more careful forward-running measurements, finding that the spin of the toy in fact depends on the geometry of the arms in ways the simple momentum argument did not predict. The UCL-NYU paper sets itself apart, its authors say, by approaching the same dependence from the opposite direction and isolating the jet path from the rotor's motion.

The research also carries an implicit caution. In an era when large language models can produce confident-looking answers to well-trodden textbook questions, the sprinkler problem has become a minor internet case study: the same question asked of several models returns several different directions of spin, none of them grounded in the actual physics. The point of running real water through real nozzles, the authors note drily, is that it occasionally tells you the textbook is wrong.

What is still uncertain

The team is careful about what the new rig does and does not settle. The forward-running sprinkler, with its spinning head, still has not been measured at the lowest flow rates that the reverse experiment can now access, because the spin itself becomes too slow to track reliably. Whether the two regimes connect smoothly, or whether the breakdown of the jet at low flow rates introduces a genuinely new behaviour, is the next question. The researchers also acknowledge that their results sit on a small number of nozzle designs and arm geometries; broader geometries may complicate the picture.

What can be said is that a plastic toy bought for a few pounds, clamped to a laboratory bench and run backwards, has produced the cleanest data yet on a puzzle that has lived in physics lecture halls since the 1980s. The next lawn it decorates will be doing slightly more work than it looks.

This publication framed the result as a continuing measurement problem rather than a settled verdict. The wire coverage in Science Daily emphasised the reversal technique; Monexus reads the same data as a reminder that long-standing textbook puzzles are usually settled by better instruments, not by louder arguments.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://en.wikipedia.org/wiki/Feynman%27s_sprinkler_problem
• https://en.wikipedia.org/wiki/Silly_sprinkler