Rontec, your post appeared while I was composing my last post. As I understand things, the changing phase of a second order filter is in opposite directions for high and low frequency sections, and is 90 degrees out of phase at the crossover frequency in each section. Changing the physical placement of the drivers relative to one another will not correct this. If the drivers were connected in the same polarity (i.e. both moved the same direction with the signal) the there would be a net 180 degree out of phase condition at the crossover frequency... perfect cancellation if the acoustic centers were aligned. Connecting the drivers in opposite polarity provides constant power response from the system through the crossover region. Unfortunately this results in a situation where, with transients, one driver sucks while the other blows... so to speak.

If we were dealing with periodic (i.e. perfectly symmetric and repeating) waveforms like sine waves, then physical placement could correct any errors. With the highly asymmetric, transient waveforms of real recorded signals, the drivers must be in the same relative polarity and aligned correctly to stand a chance of reconstructing them properly.