One experiment that was done recently (last 1/2 year or so) was to take two pairs of entangled photons, and measure one photon from each pair for polarization. These are uncorrelated. Then, the two remaining photons - one from each pair - go on after the first two are measured. Sometime later, those two photons can be operated on in such a way as to entangle them, which (retroactively) entangles the first two photons that were measured for polarization. This "swaps" the entanglement.
When the entanglement is swapped, the polarizations of the first photons will HAVE BEEN measured to correlate with each other, even though the decision as to whether to swap the entanglement as made after the measurement when the two original photons no longer exist.
Another experiment moved entanglement from one photon to the next, to the next, and then measures it. The entanglement works even though the partner hasn't existed across several transfers of entanglement.
My conclusion is that entanglement not only doesn't care about spacial separation, it doesn't care about temporal separation - which according to Einstein is kind of the same thing. The world is not what we think it is.
= new reply since forum marked as read