Nick, Troy is apparently(as Nick B suggests)using the Harman study as the source. What's being misinterpreted in some of the comments here is the "LF factor". This doesn't mean that the total low frequency energy goes down when more subs are in operation. The total actually increases, but a correction factor to "normalize" for the number of subs is applied to get the LF factor. This measures the efficiency of adding more subs in a particular configuration. In other words, each added sub doesn't increase the total low frequency output as much as would be thought simply by looking at the increase in numbers of subs.
