puzzle in I-band photometry on Data Disks 18

[Stupendous Man <richmond@a188-l009.rit.edu>]

  This has to be brief, since the power to our building is going
down in 14 minutes.  I've been looking at photometry of stars
on Data Disks 18g, 18h, 18i.  Repeated measurements of stars
in V-band (on the same night) show very low scatter, as one would
expect.  Good.  However, repeated measurements of stars in I-band
(on the same night) show a floor of about 0.03-0.05 mag.  

  At first, I thought it might be a shutter problem, but the
variations in stellar brightness from frame-to-frame are
not correlated with variations in sky brightness.  Kill that idea.

  I've now discovered that the variations depend upon the position
of a star within the frame.  Specifically, stars on the East side
of the chip tend to be brighter by about 0.10 mag than the same
stars on the West side of the chip.  I'm using aperture photometry
through a circle of radius 4 pixels.  The FWHM is around 3.5 pixels
on these I-band images.  There isn't much variation in PSF across
the frame.  

  The I-band flatfield does not show a pattern similar to this
pattern of residuals.  Hmmm.

  This sort of effect isn't uncommon with wide-field cameras.
I can take it out, once I know how the effect varies with
position.  For an example of the gory details, see a paper
by J. Manfroid, "Stellar calibration of CCD flat fielding,"
published in A&A, vol 113 (1995).

      http://adsabs.harvard.edu/cgi-bin/nph-bib_query?
          bibcode=1995A%26AS..113..587M&db_key=AST&high=3b152badf716502

  Sigh.  More work for me after I return from the AAS meeting.

  I wonder what exactly is the cause of the effect?  Could other
people please check this out and verify that I'm not making some
stupid mistake?

                                          Michael