Saibal Mitra wrote:
Now in the article, Afshar claims to have measured which slit the photon passed through and verified the existence of an interference pattern. However, this is not the case - without the wires in place to detect the presence of the interference pattern, photons arriving at detector A have passed through slit A, and vice-versa with detector B and slit B. However, with the wires in place, some photons are scattered, indeed some photons which passed through slit A will arrive at detector B. With both slits open, and the wire placed exactly at a null point of the interference pattern, the photons passing through slit A and arriving at detector B exactly counteracts the photons passing thoguh slit B that have been lost through scattering. The mathematics of quantum mechanics assures this, coincidental this may seem.
A poster on sci.physics.research elaborates on this point a little with a nice thought-experiment involving enlarging the wires until they are almost touching, at which point you just have a new set of "slits":
Now I haven't done any calculations or read the New Scientist article except looking at the lab setup graphics, but if I would hazard a quick guess, it would be that it will turn out that even if the wires are placed in the interference fields valleys, the finite width of the wires will diffract just enough photons to erase the which-way information that was gained by focusing the detectors at the holes in the wall through the lens.
Consider the limiting case with wires placed with their centres in the interference fields valleys as before, but expand their width so much that they almost touch each other. What you have now is yet another wall with a bunch of slits in! Obviously, almost all which-way information is lost after the wavefronts pass these almost infinitesimal slits since they will diffract the photons equally no matter from which hole in the *first* wall they originated, so any detector placed after this obstacle will be like running a new multiple-slit interference setup (although with the lens now severely defocusing the too-closely placed new slits). And since the which-way information from the first wall is erased, interference is free to happen between the first and the second wall. After the secondary wall the detectors can pick up which-way information causing them to behave as if there was little subsequent interference.
Conversely, the other limiting case is with no wires (or secondary wall) present. Then all which-way information is present and again the detectors behave as if there was no interference.
The experiment shows a case in between these limits and the effect I guessed at above could (and should, according to traditional QM) turn out to always cancel any attempt to find both 100% interference and 100% which-way information. This would be better showed with some calculations of course...