In ISIS/FOS spectropolarimetry, the usual method of observing will be the ``staring'' mode, described in Section 3. As mentioned above, it can be looked on as a very slow modulation method, but we wish to distinguish it from another option, which makes use of telescope movement to implement a method with somewhat faster modulation (0.05 Hz, as opposed to 0.005 or thereabouts). Such a method could be used when the analyser has only one beam (Polaroid), which may be unavoidable in some applications (e.g. when a spectrum along the full length of the slit is required at one go, rather than with the comb-type Dekker); it will be referred to as the ``modulation'' mode.
Ideally, the frequency of modulation should be of order 10 Hz. The polarization module is mechanically capable of this, but no rapid-readout detectors are available. An alternative is to use separate pixels for the 2 states of the modulator by moving the telescope to another position to expose another part of the CCD without reading it out. Flat-field information is then required to relate one part of the CCD to the other. The rate of modulation will be determined by delays in moving the telescope, Dekker and waveplate between their 2 positions; at present, a rate of 0.1 Hz is likely to be the maximum attainable. Is it worth it, and if so, under what conditions and for what projects? If not, we shall have to wait for the rapid-readout CCDs to be implemented, or the advent of other types of detectors.
Whenever the calcite slab can be used, atmospheric noise (due to scintillation, seeing or extinction variations) can be eliminated fairly effectively, and the calcite plate eliminates light loss. However, when using the calcite slab, long-slit objects have to be split into sections, to accommodate the other beam. The calcite slab may also not be fully compatible with the cross-dispersion option planned for the future (and very desirable for polarization work: at lowest dispersion, the whole wavelength range can be covered in one exposure with 2 CCDs, ideal for 10-Angstrom resolution spectropolarimetry). Similarly, FOS multi-slit or aperture-plate spectropolarimetry will have to use a Polaroid analyser (mounted in the multi-slit unit). In all those cases, a Polaroid analyser will have to be used, necessitating modulation to reduce atmospheric noise.
To avoid spurious system gain variations (causing apparent polarization) by wandering of the stellar image (spectrum) over the surface of the non-uniform detector, it will be necessary to pay special attention to slit use: the centering in the 2 states of the modulator should be the same as far as possible. On point sources the method will probably require either a very narrow, short slit (say 0.5 arcsec for 1 arcsec seeing) or a very wide one resembling a photometer aperture (say 5 arcsec for 0.5 to 1 arcsec seeing). Use of the autoguider is highly desirable in all cases.
We have tried to implement an automatic routine based on an ICL procedure with observer control of some of the parameters. These parameters are: the total number of partial exposures, the 2 waveplate positions, the Dekkers in the 2 states, the ``throw'' of the telescope position, and the length of the partial exposures, separately for the 2 CCDs. While this way of observing is essential only when using the Polaroid analyser, the routine can be used with the calcite plate, for instance when it is important that both Stokes parameters be observed as far as possible at the same time. With the polaroid analyser, one switches between 2 halfwave plate positions 45 degrees apart and deduces one Stokes parameter from this; the other Stokes parameter needs another frame. With the calcite plate, the halfwave positions may be taken 22.5 degrees apart and a second observation (with positions differing by 45 degrees from the first) is needed to correct for the gain ratios (see the section on the staring mode); the Stokes parameters are then effectively observed at the same time.
Experiments quickly showed that modulation by this method will only work well for point sources if conditions are photometric and essentially photometric practice is adopted: the focal-plane aperture should admit very nearly all of the stellar light. It seems therefore that it will be limited to the cases above and that for compensation of atmospheric noise we shall rely mostly on the calcite slab analyser.