It's possible to learn how to hear through the signal chain of early recordings--especially the "electrical process" that came in in the late 1920s--and mentally recreate the original sound. By 1938, when the Ryerson side was cut, audio quality was far better than the old play-to-the-horn recordings, and in it I can hear a characteristic "doink" that I associate with 17-inch Gibsons.
There would seem to have been as much guitar-to-guitar variety in the 30s and 40s as there is now,and the videos sampled here (or demo'd by, say, Jonathan Stout) are evidence that sweeter-voiced archtops were out there pretty much from the beginning. But I suspect that among big-band players there was some selection pressure favoring the very loud, relatively dry, midrange-honk voice that we hear on the records.
Then there's the context of many of the solos on which we base our sense of what a vintage archtop sounds like: the guitar steps out for a chorus but for the rest of the tune is in its rhythm role, which means the player was going to be using his orchestral axe. (For example, Al Casey on the small-group "Buck Jumpin'.") An interesting exercise would be to go through the solo/duo recordings (Kress, McDonough, Van Eps) to see whether those players chose guitars with sweeter voices or just stayed with their big-group instruments. I suspect they're pretty much all using their regular gigging instruments.
Eddie Lang's Gibson (a pre-Advanced L5) has a bit of that constricted voice I associate with a big-band guitar--but he still manages to get it to sing. (His hands must have been really strong.) Matt Munisteri plays a similar L5 on a promo for his lesson series here:
https://pegheadnation.com/string-sch...of-jazz-guitar
The modern recording tech lets us hear what Eddie's guitar might have sounded like without the bandpass constriction.
On edit: Carl Kress's "Sutton Mutton" shows off what a 30s archtop could sound like, even through the recording tech of the time:
https://www.youtube.com/watch?v=JJD-jdU6DTw