Imperfections in the system result in a small loss of modulated (i.e. polarization) signal. A correction for this modulation inefficiency is obtained by measuring the polarization of any star through a polarizing filter (located in the A&G main filter slide). In restricted wavelength ranges the Polaroid polarization filters produce a 99.99 % polarized beam and any deviation from that must be attributed to system inefficiencies (or to errors in the correction for scattered light !).
Two types of Polaroid filter are (or will be) in use: a UV-type 99.99 % effective from 320 to 780 nm, and a visual type from about 420 to 810 nm. Check with your support astronomer which filters are available. For the full wavelength range there is a calcite plate but this only works for point sources and in good seeing (not fully tested); one would like to have a twin-tilted-plate polarizer to produce about 5 % polarization on unpolarized stars, but this is only in the preliminary planning stage. Whatever polarizer you use for calibration of modulation efficiency, use exactly the same ISIS configuration as for your observations and correct the polarization of programme objects for the inefficiencies. Fig. 8 shows the data for a laboratory superachromatic halfwave plate; the ISIS component is similar (though not identical).
[ TIFF ]
Figure 8: Actual performance of a halfwave plate similar to that of ISIS.