These NRC archival photos are interesting, but I have never seen them before - but I was there at the claimed date . . . I suspect that the dates are wrong, as it looks as though they are testing the acoustical attenuation of ear defenders - the hard shell, liquid-filled cushion design was invented and patented by two of my colleagues before I got there in 1965. The speaker was almost certainly an AR-3, which was there when I arrived, and which I used in my first blind listening test in 1966.
One of the first things I did when I got full-time access to the chamber was to remove the acoustically hostile catwalk. The Barton photos show the new chamber wedges with the mic arrangement as it was when I left - along with the ancient, but obviously still functional, measurement apparatus. I left in 1991 - a long time ago.
For measurements below the approximately 80 Hz cutoff frequency, as quoted in post #125, calibration is necessary. The original calibration done in 1983 used measurements on a 10 meter tower as a reference. It was a significant effort, but deemed necessary. One can also use ground plane measurements. Incidentally the determining factor for low frequency cutoff is the length of the wedges - about 1/4 wavelength is the determining factor. To move the cutoff frequency down to 30 HZ would require impractically long wedges: about 10 ft! Chamber size is not critical, except that with long wedges more space is needed between them to be able to do the measurements.
The calibration woofer, as stated, was a closed box monopole. ANY bass radiator with different radiation characteristics, such as dipoles, long towers, bass reflex, cardioid, etc. will couple differently to the residual standing waves in the chamber and the calibration will be wrong, and will change with angular orientation within the standing wave pattern - exactly as happens in listening rooms. As I said, the most likely realistic response is the sound power metric, which soundstage does not purchase.