A totally different approach to sky correction from that described in Section 3.3 is to compensate it optically by allowing the e-beam from a sky aperture to coincide with the o-beam from the star and vice versa (Fig. 10). The sky contribution in both stellar spectra now contains both polarizations and the ratio obtained is the square of . Ignoring the sky spectra and just reducing the stellar spectrum as before, we obtain the uncontaminated stellar polarization, but with a scale factor of . Often such a scale error is preferable to an additive (vector) polarization error,but it can be ruinous at the wavelengths of sky emission lines. Since these probably are unpolarized, they only cause a scale error in `normal' operation, whilst the `compensation' method doubles their strength. If experiments with such automatic sky compensation are undertaken, the spacing of the Dekker apertures should be equal to the beam separation caused by the calcite plate. This is quoted by the manufacturer as: 2.57, 2.20 and 2.14 mm at 300, 633 and 1000 nm respectively. Experiments will be necessary to find the best Dekkers for this method.
[ TIFF ]
Figure 10: Optical elimination of sky polarization.
GOOD LUCK !
The appendices are intended for fluid but (we hope) up-to-date information on the polarization unit and its performance. Please communicate to us any results or considerations which you feel ought to be recorded here and tell us when the information has ceased to be useful.