[fitsbits] Representation of polarimetry data in WCS

On Wed, April 26, 2006 9:25 am, Malcolm J. Currie wrote:
My preference was not to have a Stokes axis, because each element
measures a different thing.
Whether each element measures a different thing is very much a
matter of perspective :-) Each element on the Stokes axis is a
(linear combination of) time averages of various products of
the two components of the electric field. To me that's no less
different than measurements at different frequencies/wavelength
or different positions. That's perhaps just slightly less evident
with true Stokes parameters than with the natural output of a
correlation system (XX*, XY*, YX*, YY*, which I think also
have their own Stokes code in the standard), but still easily
good enough by my standards.

This is probably more natural to a radio-astronomer since coherent
amplification allows all these parameters to be measured simultaneously,
but I'd argue that it's a very useful point of view. Of course
you might instead just conclude that radio-astronomers had undue
influence on the early development of FITS :-)