Thanks for the insightful comments everyone. This has been a good thread.

I had not considered the T(1,1)2 mode being concentric, with significant anti-phase output from the sides, but of course this makes perfect sense. I'd imagine both mass and stiffness in the rims tend to impede this output, but would affect the size of the node line ring differently.
Trevor Gore wrote:
The impedance mis-match effect and the "heavy sides" effect are different. One effect of heavy sides is that it moves the node line of the T(1,1)2 mode out towards the rim. The T(1,1)2 mode (main top mode) is interesting in that it is actually a concentric dipole, because the area outside the node line moves quite a lot, but obviously 180 degrees out of phase with the area inside the node line. So there is some phase cancellation of the net radiated sound.