What I was trying to say is that if there's an inherent problem in measuring low frequency response (and I have no theoretical basis for "if" - as far as measuring I'm just "point and click") then what apears to be a huge peak to me may in fact not be a real problem.

I may be either over compensating or misinterpreting the observation.

But it seems indisputable to me that, difficult or not, the only place worth measuring and correcting is in the actual environment.

In my real life, everybody believes that factory test data is a useful QA tool, but nobody accepts that a piece of equipment works until it's bolted down in the field and run.