Tech Note 89 and photometric error vs. position on frame

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

  I've made a first cut at determining the error in photometry
as a function of position on the CCD, using Tom's 7x7 grid
data from DS24.  You can find my results in Tech Note 89:

        http://a188-l009.rit.edu/tass/technotes/tn0089.html

  Executive summary: _if_ the effect is stable from night to night,
then we can decrease the scatter in measurements of the same star
by a factor of two (or possibly more).  

  However, I need to run many additional tests to verify that
the errors _are_ stable, and to see if there's a better way
to determine them (higher order polynomial, or modified weighting
scheme, etc.)
  
  In the best of all worlds, this effect _would_ be reasonably
stable, and we _could_ apply the corrections to data already 
stuffed into the database.  But that remains to be proved.
It could very well depend on the PSF, which might change slightly
from night to night, for example.

  Tom wrote concerning the issue of signed versus unsigned data values
in the raw images:

> Why not fix Downfits so that it writes 0 
> - 65535?  Michael, I assume that the pipeline will not care???

  Well, there's the small problem that XVista truncates all pixel
values greater than 32767 to 32767 ....  sigh.  I have been working
on this for two years now ... still not done.  I don't think it will
make any big difference -- such stars tend to be saturated anyway.

  Merry Christmas, all.

                                          Michael