It's not a digital effect: either "in phase" or "out of phase".
Essentially, "in phase" means you are getting perfect 100% constructive interference between the frequencies of the mains and the sub where they overlap (near the crossover freq). 0 deg phase diff. In other words, sound is comprised of waves at different freqs and amplitudes. When 2 signals are in phase, the peaks of one signal line up with the peaks of the other, at the same freqs. The 2 signals add together so that 2 + 2 = 4.
When something is 100% "out of phase", 180 deg out, the peaks of 1 signal perfectly line up with the valleys of the other. If you add peaks and valleys, typically you don't get much of anything. They cancel. 2 + -2 = 0. Hence the bass level is lowered near the crossover freqs. (A full wavelength is 360 deg.)
So if you can imagine 1 wave passing over another, the phase can be anywhere from 0 deg (perfectly in phase) to 180 deg (perfectly out of phase) back to 360 deg as the waves start to line up again, but where the peaks and peaks (or peaks and valleys) don't perfectly line up, for example, you can be 57 deg, or 235 deg out of phase.
What a lot of people don't understand is, if you are 360 deg out of phase, you are actually back in phase, but now one signal is 1 full wavelength after the other (or delayed from the other). So now ever though you are in phase, bam, your time alignment is now hosed.
Hopefully that helps!
(Much easier to illustrate with actual waves on 2 different pieces of paper.)
[This message has been edited by Kevin C Brown (edited January 24, 2003).]