Stefanl: I'm sorry, but to me that story is likely mostly apocryphal. There were no monitors Emerick could have been using that would reproduce 54KHz flat even if he could hear that high (which I don't believe), and I doubt the response of his board was extended flat beyond that frequency either. If the story has a germ of basis in fact, I would have to assume that whatever was the fault in the board (a ringing spike caused by incipient oscillation? - certainly not a -3dB dip), it must have been precipitating some audible artifacts affecting frequencies down much closer to the range of normal human hearing.
Eldartford: I can understand that smooth ultrasonic extension will aid in maintaining waveform linearity lower in the audioband, and that glitch-free gradual ultrasonic roll-off beginning well above the audioband is probably never a bad quality to cultivate everywhere throughout the recording/reproduction chain (even if a lot of what goes on up there is probably noise). But to me it seems that when it comes time for such a signal to exit a speaker, most conventional designs will be capable of only such limited dispersion that one would have to listen perfectly on-axis with their head in a vise to receive much of the theoretical benefit...
Eldartford: I can understand that smooth ultrasonic extension will aid in maintaining waveform linearity lower in the audioband, and that glitch-free gradual ultrasonic roll-off beginning well above the audioband is probably never a bad quality to cultivate everywhere throughout the recording/reproduction chain (even if a lot of what goes on up there is probably noise). But to me it seems that when it comes time for such a signal to exit a speaker, most conventional designs will be capable of only such limited dispersion that one would have to listen perfectly on-axis with their head in a vise to receive much of the theoretical benefit...