Per-chord is what I meant, not per-key. 3:2 means you can't have a coherent set for all tones in a key, full stop. But you could slightly shift the frequencie

Per-chord is what I meant, not per-key. 3:2 means you can't have a coherent set for all tones in a key, full stop. But you could slightly shift the frequencies of the notes in a given chord to make that chord perfectly tempered. There'd be a million details; making sure the melodic voice, bass, or both avoided wolf intervals; figuring out what to when some voices are moving over a sustained chord in others; keeping the main anchor frequency from drifting "too far" from the original (and probably determining "too far" in the first place); figuring out whether the resulting sound was going to be anywhere near distinguishable enough to be worth all the effort of hacking through all this rigamarole.