So far, all simulations have been done with a single dithering pattern that does a reasonable job of partial pixel spacing for the 6 overlapping frames within a scan. A comparison of the effects on photometry and astrometry when chosing a ``good'' (0.15,42.23) or ``poor'' (0.23,42.01) spacing was made. The latter spacing was chosen because it is representative of some of the scans made through M92 in the 6/1/94 data.
Simulations were made at b=25, using a PSF with FWHM = 2.5 arcsec. Subpixel response was included, but no dead pixels. Figures 14 and 15 show the results of the reduction. Please note that these comparisons have less data per magnitude bin than others in this memo; thus the sigmas, espcially at K < 12, may be affected by number statistics.
The photometry and positions are significantly worse in the ``poor'' dithering case. Actually, worse than we would have expected from the 6/1/94 data. It may be that, because of the ``perfection'' of our simulations (e.g. no jitter/ error in the X,Y positioning of frames), we may be overly sensitive to the uniformity of the scan parameters. A future refinement of the simulations will make the true offsets less uniform. Nevertheless, this test points out the importance of sampling the PSF over a good range of subpixel spacing.