If your crossover has separately adjustable crossover frequencies like the Bryston, you can probably get away with not removing the crossover although the net effect will not be as good as removing the crossovers, it will preserve the resale value. If not you need to remove the crossover.
If your crossover does have separately adjustable crossover frequencies, you can set the high frequencies to extend below the crossover point and the low frquencies to extend above the crossover point and let the passive crossover filter out the portion of the frequency spectrum above or below it's crossover point. The advantage is that the amplifier channels will no longer see the portion of the signal above or below their crossover points. The disadvantage is that you still have the passive crossover in there sorting out the frequencies around the crossover point.
To answer your question about the effect of leaving both in, think of it in these terms. Let's say you both have an active and a passive crossover which both attenuate the signal at 6 dB per octave (just to make up a number). The passive crossover was designed so that with it's 6 dB per octave the speakers would blend together and give you a flat frequency response. With the two crossovers in the picture you now have a 12 dB per octave rolloff. Looking at the situation from the woofers point of view, originally it was counting on the help of the midrange speaker to produce part of the frequencies below the crossover point. Since these frequencies are being rolled off more quickly that the designer intended, less of those frequencies are being produced by the midrange than required for flat response (assuming it was flat to begin with). The same thing applies in reverse to the upper end. The net result is a reduction of output around the crossover point.
Greg
