... observation

Appendices are only available in electronic form at http://www.edpsciences.org

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... self-calibration

Paper III (Hamaker & Bregman 1996) considers the matter of Stokes-parameter definitions which is of no relevance here.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... self-calibration

One may question this assumption for antennas with different feed orientations as appear in the simulation to be described. I assume here that the primary beam is the product of independent scalar effects of a circularly symmetric (pair of) reflector(s) (which is the same for all antennas) and the polarimetric effects of the feeds.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... italicized

Note the difference between Stokes parameters U,V and visibility/coherency coordinates u,v.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... form

The sign of V is given here as in Paper IV. R.J. Nijboer (priv. comm.) has noted that it deviates from the definitions given in Papers I and III (Hamaker & Bregman 1996). This is an unfortunate mistake, but it is immaterial in the present context.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... multiplier

As in Paper IV, we ignore the loss of absolute position in the Sky that self-calibration also entails.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... intensity

Recent work has revealed the existence of extended structures that may appear in interferometry to be more than 100% polarized; for such situations not only our CLEAN method, but the entire selfcal paradigm must be reconsidered.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... algorithm

S.J. Bhatnagar (priv. comm.) observed the same effect when using an entirely different algorithm in AIPS++. This strongly suggests that this behaviour is due to the nature of the problem rather than to a peculiarity of either algorithm.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... errors

This is similar to although not the same as the separate self-calibration of the X and Y or L and R channels with which quasi-scalar polarization processing starts up (see Paper II).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...$\vec{p}_{{\rm lin}}$

This does not entirely rule out the possibility that one flips the entire QUV space "upside down''. This possibility is in fact inherent in any method that relies on zeroing V, e.g. the XY phase-difference determination in the quasi-scalar calibration of linearly polarized arrays (Paper II). This seems never to have happened in practice, which shows that the usual pre-calibration is an effective prophylactic.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... polarimetry

In the past, experienced polarimetrists have occasionally derived crude "manual'' Faraday alignment and polvector-leveling corrections by visual inspection of their quasi-scalarly self-calibrated Sky images (A. G. de Bruyn, private comm.).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...k

This use of the approximation is quite peripheral; with proper pre-calibration, the errors that it might introduce are negligible. Moreover, they affect the entire Sky rather than each source on its own, and can therefore be corrected later.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.