As mentioned in the ESPaDOnS FAQ, doing continuum polarization with this instrument is not obvious.
On September 7/8 2005, tests were performed on Gamma Cas to try to answer these 2 questions:
The data consisted in very short exposures (5sec) to measure Stokes V (circular) and QU (linear). The first test was done with VQU measured while the star was well centered in the Star Hole, with a second series of measurements with the star halfway out of the Star Hole. The second test consisted of 12 consecutive VQU measurements, with the star well centered in the hole, and guiding.
Weirdly enough, when the star in sitting on the edge of the hole, its polarization is less than 1% in all Stokes parameters. When the star is in the hole and we guide, Stokes Q or U can be as high as 4%, while Stokes V (circular polarization) is around zero. The continuum polarization is not flat as a function of wavelength either. Here are PDF plots, with Intensity (bottom plot), Polarization (whatever Stokes parameter was measured, in percent), and Resampled polarization (in 50nm bins):
Star location | Stokes parameter measured | Pdf plot |
---|---|---|
IN the hole | V (circular) | |
IN the hole | Q (linear) | |
IN the hole | U (linear) | |
OUT of the hole | V (circular) | |
OUT of the hole | Q (linear) | |
OUT of the hole | U (linear) |
The second test showed that the continuum polarization varies wildly, in all Stokes parameters, from 1-2% up to almost 10%; that's a lot! Also, this effect is seen for wavelengths above ~540nm; under ~540nm, the variations are less, under about 1-2%.
Here are the results plotted:
The conclusion at this point is that for very short exposures (5sec), the precision is no better than 10%. The accuracy might be better for longer exposures (30sec) because whatever effects will average out somewhat, but this remains to be tested.