In his paper, Strekalov, D. V., Sergienko, A. V., Klyshko, D. N., & Shih, Y. H. (1995). Observation of two-photon “ghost” interference and diffraction. Physical review letters, 74(18), 3600., they state the following:
Furthermore, if $D_1$ is moved to an unsymmetrical point, which results in unequal distances to the two slits, the interference-diffraction pattern is observed to be simply shifted from the current symmetrical position to one side of $x_2$. We still observe the same interference pattern in the coincidences (same period, shape, and counting rate), except for a phase shift.
I find this statement completely strange. Why is the interference pattern shifting? If $D_1$ is moved I would just expect to have the same interference pattern unshifted but at a lower counting rate since in the end you are just moving the Bucket detector to a region with lower probability of detecting the signal photon. But no! when moving the detector you somehow introduce a phase shifting in the slits just like the one you get when you separate the slits in the propagating direction.
Even more striking is the fact that you still have coincidences even placing the detector $D_1$ in a position where supposedly the probability of detecting a photon is zero.
I hope I'm being clear enough, and probably I'm missing something related to the correlation of the photons but it is still very strange.