Wayne,

The Lexicon system does not do any averaging of the 4 mic inputs. It's inventor, Dr. James Muller, has made that very clear. If one mic picked up a dip and the other mic picked up a similarly sized peak, averaging the two results may cancel each other out and give the (mistaken) impression that there is no problem. The data correlation being used is more sophisticated than that.

What you're doing with EQ and what Meridain/Lexicon are doing with their EQ systems are completely different. For example: if those systems detected a hump at 70Hz, they would do absolutely nothing about it. They aren't looking for nor correcting anything in the amplitude domain. Instead they measure in the time domain, looking specifically for low frequencies that linger for the longest amount of time, obscuring details of sounds that immediately follow (irrespective of their frequency).

If a frequency matches a room dimension such that it keeps bouncing back and forth between two walls for longer than other frequencies do, then no seat placed between those two walls will be imune to that frequency ringing. So while levels can vary widely from seat to seat, ringing is more consistent in all seats. Reduce its decay time and you'll improve clarity across the entire frequency range, at all seats. Difficult, if not impossible, to get that sort seat-to-seat consistency when correcting for varying amplitude instead of varying decay times.

Keep in mind that it is difficult to measure decay times by looking at amplitude peaks, the way you are. That's because a ringing frequency may not be louder than other frequencies, so it will measure flat in amplitude but still have a long decay time. Other times, a peak is a symptom of a ringing frequency. In those cases, when you fix the problem (long decay time) you end up fixing the symptom (amplitude peak). So, even though they aren't attempting to do so, the Meridian/Lexicon systems can sometimes end up inadvertantly smoothening and flattening some of the frequency response.

Make sense?