Summary of relative gain matrices (rgm) computed using sharcsolve and
scanning/no-chop observing mode [2006-Jan-25, J. Vaillancourt]

All 384 channels:
-----------------
- 13 scanned RGM files taken and reduced using sharcsolve

- Calculated median gain for each pixel, plotted histogram of relative
difference between each pixel and its median for each file
	   - 1 file had histogram with peak significantly different from zero
	   - 1 file had dispersion much larger than other files
	   - 1 file had dispersion moderately larger than other files
	   - remaining 10 files had peaks ~ zero, stddev ~ 1 - 6%
	   - Possible trend in stddev with atm. attenuation/loading
             (stddev decreased with increasing loading)
	   - No trend in mean deviation from median with atm. loading

V-array only:
-------------
Same analysis as above, but keep all 13 RGM files
     	      - all histograms have peaks ~ zero, stddev ~ 0.4 - 3 %
	      - if remove 3 files mentioned above, stddev ~ 0.4 - 2 % 
	      - No trend in stddev with atm. loading
	      - Possible trend in mean deviation with loading (mean
                increases with loading)

H-array only:
-------------
Same analysis as above, but keep all 13 RGM files
     	      - all histograms have peaks ~ zero, stddev ~ 1 - 8 %
	      - if remove 3 files mentioned above, stddev ~ 1 - 8 % 
	      - Possible trend in stddev with atm. attenuation/loading
              	(stddev decreased with increasing loading)
	      - No trend in mean deviation from median with atm. loading

Comparison of H/V only with all 384 channels yields the same rgm to within 1%.