I have a dip/suckout in the 2251J cabinets at 800 Hz. Both cabinets do this and while the room is an influence, nearfield measurements would indicate something closer to home. I had checked with the published JBL spectrum and found no dip at this point in the driver's response.
I decided that this was potentially due to a cancellation of the wavelength due to the reflecting wave in the cabinet cancelling out the motion of the cone. There is no padding in the cabinet except for a piece of felt glued to the back wall. This is the nearfield spectrum with the driver through the crossover. Rob had mentioned that he thought the speakers were a bit bright in the upper midrange. This is probably what he was hearing. The dotted lines represent increments of 2.5 dB.
I theorized that by moving the back wall forward, but not taking up a lot of cabinet volume could potentially move the null up the scale because this represents a shorter wavelength and higher frequency on the bounce. Also, if the material were "soft" it would absorb some of this energy reducing the bounce.
I cut a couple pieces of 1-1/2" thick styrofoam and made them a tight fit on one of the the cabinets with the cut out for the port. This has to be two pieces to get them into the cabinet as the back is sealed. I put the speaker back in and everything else is just as it was in the first test. This is the result of adding the styrofoam. The lower end has come up considerably, the 800 Hz dip is reduced a bit, and the 1kHz+ peak is removed leveling this part of the spectrum to that below it. (There you go Rob.) The whole band is considerably smoother.
I will try removing the styrofoam and see if I can get similar/better results using some pink insulation sitting in the loft. The styrofoam is much harder to get the cabinet pattern right.