Flipping the sprinkler uncovers a long-standing puzzle in fluid dynamics
Mathematicians in the UK ran a lawn sprinkler in reverse to probe a paradox that has nagged researchers for decades, and came away with new evidence that the leading theory misses a piece of the story.
On 13 July 2026, researchers at the University of Oxford published a study in Physical Review Letters that, on the face of it, asks an almost childlike question: what happens if you pump water into a lawn sprinkler instead of letting water pump out? The twist, the authors say, is that the answer exposes a real gap in the textbook treatment of the humble device, a device whose behaviour has been debated in the open literature since the 1940s.
The "silly sprinkler" problem has long been a foil for physicists. Push water through the curved arms of a rotating sprinkler and the nozzle tips spin in one direction. Suck the water back out and intuition says the arms should spin the opposite way, like a turbine in reverse. The puzzle is that, under some conditions, they do not. The sprinkler sometimes sits still or even keeps turning the same way, defying the neat momentum-balance argument that students have been taught for generations.
Oxford's group, led by mathematicians at the Mathematical Institute, set up a transparent acrylic lawn sprinkler inside a water-filled tank, then filmed the system with high-speed cameras as they reversed the flow. By submerging the whole device, the team removed the messy contribution of the surrounding air, the layer of moving fluid that is genuinely hard to control in a garden. That left them with what is, in effect, the cleanest possible test of a textbook assumption.
Their result, captured in slow-motion video and posted alongside the paper, was striking. Even with the air removed, the reversed sprinkler did not behave the way the standard equation predicts. The arms still turned, in the same sense as before, at low flow rates. The discrepancy only disappeared at higher flow rates, when the jet of water leaving the nozzle was energetic enough to dominate the dynamics.
The team's reading is that the existing model treats only the momentum carried away by the water jet and ignores the suction effect, the way the moving water pulls surrounding fluid along with it, creating a flow inside the sprinkler housing itself. That second contribution, often called the Coffman effect after one of the first modern attempts to catalogue it, can flip the sign of the torque and cancel out the jet. Once the jet grows strong enough, it overwhelms the suction, and the textbook picture returns.
This matters well beyond a garden ornament. The geometry of a rotating nozzle is the geometry of any number of industrial systems: turbochargers, jet-engine starter turbines, agricultural spray rigs, even the small reaction wheels that orient satellites. Each of those is designed against the same momentum-balance equations that Oxford now says need a second term.
For a field that prizes itself on closing its own loopholes, the paper is a quiet but pointed demonstration that old problems are not necessarily solved problems. The leading theory for the sprinkler was first written down in a serious form in 1945, when the physicist Richard B. Chapman submitted a short note to the American Journal of Physics arguing that reversal should produce a force in the opposite sense. Eight decades later, with high-speed cameras and a controlled tank, Chapman's intuition is half-right and half-wrong at the same time, and the resolution lives in a region most textbooks do not draw.
None of this means the textbooks are about to be rewritten. The conventional jet-momentum calculation is fine for high-speed devices, which is most of what engineers actually design. What changes is the boundary line: the paper effectively redraws the threshold at which the simple picture works, and gives researchers a clean experimental handle, the reversed flow, for separating the jet contribution from everything else.
The Oxford group has released its video data alongside the paper, an unusually open move in a corner of fluid mechanics where bespoke rigs are usually guarded. That choice is likely to invite replication in other labs, and possibly fresh attempts to fold the Coffman effect into a single closed-form expression that designers can plug into a spreadsheet. Whether anyone manages that is the kind of question that tends to sit unanswered for a few more summers.
What is now reasonably clear is this. The century-old sprinkler is not a toy. Run it backwards, in a tank, and it tells you that a chapter of classical mechanics that has looked settled since the 1940s still has a missing page. The job of writing it is, by all evidence, just beginning.
Desk note: Monexus framed this as a working scientific paper with direct engineering stakes, not as a curiosity piece. The reverse-flow experiment is the reportable news; the broader theoretical debate is context, not the lede.
Wire provenance
This editorial synthesis draws on the following public wire/social posts:
- https://en.wikipedia.org/wiki/Feynman_sprinkler
- https://en.wikipedia.org/wiki/University_of_Oxford_Mathematical_Institute