I did some SPL measurements a couple of days ago, and the results are not conclusive; that's why I was looking for technical theory. With 25Hz warble tones played back at 82db measured at my listening seat, there was approximately a 1 db difference in sound level, as measured at my listening seat, in favor of no grille. At 20Hz the difference was nearly 2db. At 31.5 Hz I must have a room mode, because fine positioning of the microphone changed the measurements enough I couldn't tell what I was measuring. At 40Hz the difference was about 0.7db.
When I replayed some of the tracks I thought "sounded" better, it was mostly the feeling of a bass "shock wave" that made me think there was a difference. I'm not sure if I could tell blind-folded, but I'd probably bet against myself unless the bet was under $10.
I haven't run bass distortion tests yet.
Overall, the results still seem to support leaving the grille off. The measurements are repeatable and consistent, but, who knows? I've noticed at these frequencies another person in the room, or my position in the room, affects the measurements.