Eric, I didn't mean that the other approach was being done in FITS,
I know it was done in some non-FITS astronomy data packages (if I remember
correctly, the ancient Einstein Observatory data system was an
example). I agree that there is only one right way to do it in FITS,
and am merely advocating that we guard against any misinterpretation.
I don't recall any cases in which this has actually been done wrong
with FITS files, although I did have to stomp on some programmers who were
trying to once...
Jonathan