Gordon, these charts alone are not very helpful.
If the subs are identical, they will have identical GD of the direct sound and the trace will smoothly increase at the lower frequencies. When they are all timed together, they will retain that same GD of the direct sound. If they are time a little differently, there may be some small impact, but I would expect it to be insignificant. The larger the timing difference the larger the impact.
Anything that impacts the SPL response will impact the phase and GD. The roughness variation of the posted curves is the result of the SWs different interaction with each other and the room. Room effects are large when measuring at the LP. It is a major contributor to large variation of the SPL and thus phase and GD. Deep nulls thus result in major phase and GD variation at that frequency. EQ impacts the results for the same reason.
All the old advice still applies. Place the SWs and LP to minimize the raw SPL variation at the LP. Treat the room and apply EQ to further smooth the response. Avoid EQ to deep nulls.
When timing the mains to SWs XO handoff, a timing error can have a major impact to the SPL in the XO region. There are 2 or 3 options that result similar favorable SPL response but different phase and GD responses. This choice may be more likely to result in a different bass sound quality.
