A full polarization measurement takes four exposures. If the source is
variable (in polarization) within the time it takes to complete one
full polarization measurement, the results will come out wrong. If,
however, the instrumental gain ratio
is known and stable in time (see Section 3.2), two exposures (with the
halfwave plate at 0 and 22.5 degrees) are sufficient to derive the
polarization vector. The gain ratio can be readily derived from a full
polarization measurement (i.e. four settings of the halfwave plate) of
a constant source. Tests have shown that this gain ratio (averaged
over the full Dekker aperture) was stable to within 1 % during a
night, except in the cutoff region of the dichroic. Do not expect the
gain ratio to be the same after moving any of the optics, since the
throughput of many of the optical components depends strongly on
polarization and the beams within the spectrograph are highly
polarized.