Mass is not reduced, it is all on the same sheet of Mylar. Is there any divider between the bass section and the tweeter? The high frequency dispersion will be improved and that can make a difference.
Yes, they did sell speakers that needed a separate bass system. I have the Tympani IIIA-T/M, the tweeter/mid of the IIIA. They play down to about 65-70 Hz (lowest tuning frequency on par with the MG 1.6). Sure the tweeter and mid should sit far closer to each other. I modified mine many years ago and that made the "wide mouth" disappear. The MG 10 was also a tweeter/mid speaker.
Still, I do not find the size of the MG 20.7 to be much different to the MG 3.7. The Tympanis are very much different in size....
As I see it, mass is reduced at the highest frequencies, where it matters, because those frequencies are delivered via the .5-way crossover only to the single-loop supertweeter section. So nominally, only that part of the diaphragm is driven at those frequencies, and only those two foil traces, and the mass is only 1/4 that of the entire tweeter/supertweeter (if I'm correct in remembering that there were 8 traces in all).
In practice, the supertweeter segment is coupled via the Mylar sheet to the adjoining tweeter Mylar. However, electrical damping through the 4 ohm voice coil should present a high acoustical impedance to the supertweeter segment, acting as a partly-effective edge clamp. (This won't occur in the other direction, since the crossover is .5-way and the supertweeter segment is driven along with the tweeter segment in the upper midrange; the mass in that range won't be reduced.)
You can see in BigguyinATL's imaging of the MMG diaphragm that the tweeter energy is pretty successfully concentrated in the tweeter area. That more energy is successfully delivered to the supertweeter is also evident in the fact that it improves high frequency dispersion. If that weren't the case, the supertweeter would be sonically useless, even detrimental (since it reduces impedance at the very top).
I think they took measures to increase the bass output of the 20.x, and they certainly lowered the resonance (though that could also reduce maximum output by increasing Xmax). I'm stil not sure what the "diplanar" technology is, whether they're forex isobaric, with double diaphragms. It apparently doesn't mean that they're push-pull since the DWM is also a "diplanar" speaker but has a single-ended magnet assembly. But note that the DWM has *two* independent voice coils and each covers the entire surface of the speaker. (Other possibilities though including side-by-side wires -- how would the amp react to the inductive coupling between channels? -- or wires on both sides, which would have the same possible pitfall as well as making the diaphragm less limber.)
And did they increase the magnet spacing and Xmax? The second magnet assembly in the 20.x increases B, meaning that magnet spacing/Xmax could in theory be increased without losing sensitivity. And their sensitivity is about the same as the others, whereas it would presumably have increased if they'd merely doubled the magnets.
So I'm not sure how it stacks up, objectively, to the bass of a Tympani, particularly since most of my experience with the latter was with the 1-D's, which had a 40 Hz -3 dB point that could have given them even higher output. And I've never seen any actual comparative measurements of e.g., the onset of compression (though that wouldn't tell the *whole* story, since a larger baffle will couple differently to a listening room, e.g., come closer to reinforcing boundary surfaces).
