Okay, Rich, I did the previous analysis in my head, but this one is much more complex, so I used a special computer program which I'd written a couple of years ago to solve room response problems. I fed all data in and noted that the analysis was taking an unusually long time, but the result finally popped up on the screen: "This room sucks, although not as badly as that of the unit Mark S. Johnson".
Well, so much for what passes(conceivably)for incisive humor on audio boards. You've got the math down cold for the surface bounce cancellation problem, but this would appear to be a different category of interference. As you mentioned, this falls in the crossover region(Alan has reported it as 2700Hz)and a phenomenon referred to as "lobing" between the drivers results in an off-axis dip( received directly and/or by way of reflection from room surfaces)in that region. This is especially the case when speakers which would do well when positioned vertically are instead aligned horizontally. Test reports online which show off-axis curves for horizontal centers often illustrate such a crossover dip. The solution here, as in the previous problem, is repositioning(vertical here), but that may not be acceptable.
