On an Lpad, there are two floating wipers, but only one is shown in the schematics. We are using the other and it goes from ~11 to ~36 ohms before it opens completely. The circuit works best above ~18 ohms, hence R2, with ~20 ohms being near optimal.
If you saw the frequency response of the JBL 2251J, the circuit would make more sense.
I want to roll it off at about 2,500 Hz but that peak at ~3kHz makes this near impossible to do cleanly. I had to come up with some very imaginative thinking and a bit of luck from experimentation to make this work!
What I do is to roll it off a bit earlier by "over capacitating" it. If you use too much capacitance in a second order low pass filter, you will greatly increase the slope but also create a hump just below the roll off point. The more capacitance, the bigger and lower the hump. The shape of the rising hump is obvious with the Lpad is turned full up (open).
But I also let some of the signal pass through R1 or R2/VR1. This portion of the signal is NOT IN PHASE with the hump and the more you let through, the lower the hump becomes until you have a fairly smooth response up to the ~2,500 Hz roll off point. VR1 lets you selectively tune the hump.