Ed, better to replace the pot with fixed resistors once you determine the correct level. This is best done with measurements but pink noise, adjusted to sound the same on left and right channels, is also a good alternative. The latter does require trial and error and learning what pink noise should sound like.
The ACOUSTIC measurements you show above have too much high frequency for most listeners. You want the response to be FLAT but only on the tweeter axis at 1 M from the speaker. It is important to take this in a precise way, that is within 1/2" of identical mic and speaker positioning. ?At the typical listening position this will translate into about 5dB down at 15khz due to in room power response (the total energy radiated into the room vs on axis)?
The fixed resistors will both sound better and more importantly be stable. The variable pots are subject to wear and degradation. See post 6307 for a method to do this. Based on you ACOUSTIC measurements, the only ones that matter due to driver variations, the midrange to lower pot settings on the functioning speaker are the ones to shoot for. Using pink noise is a VERY SENSITIVE test, far more sensitive than music. The correct way to do this is to have helper adjust the pot while you listen what you are shooting for is the smoothest sound with no sudden "peaks" in the noise character. With a little practice this will produce a level accurate to about .5 dB which is excellent. If you have a real time spectrum analyzer function of your software AND you use (1/3) octave based measurements you can use this to calibrate your ears to hear these fine changes.
Charlie Conger