BoB-

1) As flat as I could get them by eq'ing out room induced peaks. Know how I knew there weren't any 6 dB rises due to the interaction at the crossover point? Because there weren't any major differences between sub alone, vs mains alone, vs sub+mains. Just the rises and falls due to the slopes on the filters applied to each. (Really neat to see in the real world, btw.)

2) I averaged the responses from where I sit, and the 2 other best listening positions. (Yes, I know what I'm doing.)

3) Very accidentally, I have a "good" room for acoustics. One side wall is open to the kitchen, so no reflections there. The other side wall is dominated by a sliding glass door, with a curtain in front of it. Carpeted floor. I even went so far as to put a make shift absorber in the corner of the room with the sidewall. One of those cardboard fabric bolts, covered by a few layers of fabric, propped in the bottom of the corner.

4) RS meter, every Hz from 20 to 98 Hz. With correction. Although to be honest, all the correction does is to change the tilt of the readings, doesn't hide any strange perturbations in the measurements.

Maybe instead of spending so much time trying to model these things with software, you should get out and do some real world measurements...