I think we should post the average of all of the speakers. It gives a context for the effect the room has on the responses of each. It is clear from that average that much of the variance that might be of concern in each speaker is due to the room.
This is the curve we get by averaging together all of the final measurement data (six per model, 30 total) from final speaker locations with various amounts of smoothing. It shows the typical trends we heard, although there were exceptions to some of these trends with certain models.
All that is worth noting in the overly-detailed 48th-octave smoothed plot is the sharp peaks and dips due to room modes, all verified by Room EQ Wizard's Room Analysis function. This was expected, since our speaker placement made no effort to reduce these effects. We may make efforts in the future to reduce room mode effects, but not if it hurts our soundstage very much.
The one-octave and third-octave smoothed plots show an overall emphasis trend below 500 Hz. This says that all of the models tested probably had some bass emphasis built in by design.
All the plots show that none of the models could produce bass below the 48 Hz room mode peak, and some should have been able to. Bass traps working overtime? Perhaps. Another area to investigate.
The little emphasis area at 1.4 and 2.2 kHz on the third- and twelfth-octave smoothed plots were probably designed in. These are common frequency ranges for slightly emphasizing detail and presence.
The rolloff above 5 kHz in the one-octave smoothed plot resulted from our off-axis aiming to get soundstage depth.