Peter,
i think the experiment i spoke of was first meant as a thought experiment, figured out by Zeilinger and Mike Horne in 1985 and then presented at a conference in 1987, as far as i remember. It indeed was not a quantum eraser experiment, since i confused it with another experiment. So here it is:
Take a source that produces twin-particles which propagate in opposite directions. At each direction from the source, place a double-slit barrrier. The source and the distances between two slits on a barrier is such that at the measurement screen behind each double-slit, there does not appear an interference pattern, but a homogenous gray.
Say, the twin-particle propagating to the left is called A, the other, propagating to the right, is called B.
Scanning the detection plane (with a detector B) of particle B (from left to right or vice versa) gives a random pattern of impacts which lead to the homogenous gray.
But repeat the experiment with the difference that now we also install a detector (photomultiplier or such) in the detection plane for particle A at a specific place. Now repeat the experiment by measuring the relative frequencies of impacts at the measurement plane B (while detector A is always at the same location) by again moving it (from right to left or vice versa). Now you get non-random relative frequencies of the impacts, they form an interference pattern.
For the relevant coincidence counts, detectors A and B together produce an intererence pattern in the plane of particle B. One can also once more change the relative frequencies of impacts at the detection plane of particle B by slightly shifting detector A. One can do this until the maxima and minima at side B for the placement of detector A in the first run and in the second run (the latter the one with the slightly shifted detector position) build up together a totally washed out 'interference pattern'.
The fact that a detector is used at side A alters the relative frequencies of impacts on side B, irrespectively of how large the distances between the two detector planes are. Since nothing has changed at the source and at side B, one is forced to correlate the initial change of patterns with the use of detector A. But how can the use of detector A influence what detector B measures...
My term 'Bell deniers' is confusing. I should have written entanglement deniers, since you accept Bell's theorem, but do not think that spooky-action-at-a-distance is reality, but only a misunderstanding. In a certain sense i do so, because i do not think that there is really an action at a distance in space and time. But i rationally think that there is a reason for the above described behaviour, although maybe not tracable with our usual notion of causality.
The above experiment was conducted and confirmed. In this experiment one can neglect the particles which did not hit the detectors or did not hit the double slits.
Best wishes,
Stefan Weckbach