Ideally, we would like to have at our disposal a "universal" instrument which could record all available information, i.e. the full Stokes vectors I, Q, U, V as function of spatial coordinates X and Y and wavelength A at given time t. The practical limitations of spectral dispersers now being used for high resolution observations, narrow band filters or spectrographs each permit only the measurement of a limited sub-set of these parameters. Filters record I(?,Y) at a fixed wavelength and the spectrograph record (I,?,Y) at a fixed X, where X is the direction perpendicular to the slit and Y along it. An important element to any instrumentation is the detector system and to achieve the above, a detector system must possess 2-dimensional characteristics. The application of CCDs to stellar polarimetry has been slow, one problem being that the polarimetric modulators favoured with photomultiplier detection are not immediately applicable. Since, instrumental, observational and data reduction techniques are very different in these two detector systems. At the start of this work (1991) high-precision polarimetric data are obtained by po-larimeters which employ photomultiplier detectors, with measurement accuracy controlled by photon counting statistics. For low levels of polarization, simple error analysis provides uncertainty estimates for q and u of sigmaq = sigmau =(2/N).