The continuous rotation of the reverse sprinkler has been a puzzle for over two decades. This article presents a series of experiments that demonstrate that a properly designed reverse sprinkler experiences no steady-state torque and does not rotate. Ignoring transients when the flow starts and stops, if any sustained rotation of the reverse sprinkler occurs, it is because a force couple produces a torque accompanied by vortex flow inside the body of the sprinkler. No steady-state rotation occurs if the vortex is suppressed or prevented from forming in the first place. Demonstrative proof is given that an ideal reverse sprinkler does not rotate.
References
The rotational inertia of the 1-quart yogurt container plus copper elbows and washer (for additional ballast) was approximately 1.71 × 10−4 kg•m2. The rotational inertia of ½ liter of water with a 5-cm radius was calculated to be 6.25 × 10−4 kg•m2. The rotational inertia of the deli container air sprinkler was approximately 4.15 × 10−5 kg•m2 (including the turntable doubles this). Treating the air inside as a rotating disk, its rotational inertia is approximately 4.71 × 10−10 kg•m2.
Jenkins (Ref. 7, p. 1278) points out that having the water acquire a net angular momentum around the sprinkler pivot in the absence of an external torque seems like a violation of Newton's laws only because we are neglecting the movement of the tank itself. Likewise, we argue that since the wall is rooted to the Earth, the Earth moves to conserve angular momentum.
This information is the result of an informal poll taken on the tap-l listserve. The University of Utah and the U.S. Naval Academy have the soda-can-in-water versions.