[evlatests] More on 'closure' corrections

[Rick Perley]

    I reported earlier that an image of my target source (a 
well-resolved, strong calibrator), made using 'closure corrrections' 
established from an unresolved calibrator, improved after using 
'self-generated' closure corrections. 
    We don't expect this should be necessary, as the origin of these 
corrections is in the phase hook of the VLA's bandpasses, which should 
know nothing about which source we're looking at.
    I returned to the data, and noted that the errors could also be due 
to a faulty, and/or incomplete, self-calibration process.  I then 
restarted this from scratch, taking more care to ensure non-physical 
components did not enter the models utilized.
    The result of this is that a much better image was generated, 
without resorting to 'self-generated' closure corrections.  The image is 
far from perfect, but I can plausibly (I claim!) ascribe these to the 
short observing time span (2 hours), combined with a  complicated source 
structure. 
    My current conclusion is that the established closure-correction 
procedure, using a clean point-source calibrator, is sufficient.