In order to create a successful calibration technique and understand data quality assurance, it was necessary to carefully examine MkIII data. A meeting was conducted to review data spanning the lifetime of the MkIII. Comparison of MkIII data taken since 1980 confirmed the overall consistent quality of the raw data. Problem time periods could be traced to volcanic eruptions, and one period of poor optical alignment.
The primary ongoing problem was calibration.
Due to systematic errors in the original calibration technique,
there were negative intensity regions near the poles and
excessively bright regions near the
equator. Some of these errors were attributed to
telescope polarization at the first
objective lens, 
, since
orientation of the polarization pattern rotated when the 
was rotated.
The calibration opal also had systematic polarization
which needed to be accounted
for in the calibration.
After accounting for all instrumental polarization, there
remained the possibility of
residual polarization due to
the sky.
Concentric rings and radial rays were in the calibrated images.
The intensity images often showed one pole brighter than
the other with peaks of intensity at some
scan azimuths.
There was a systematic azimuthal offset between clockwise
(CW) and counterclockwise (CCW) scans.
Hardware and software changes were implemented to improve data quality.
The gain of the intensity channel was increased from .25
to 1.0 to reduce digitization level
limitations.
Similarly, the gain of the polarization channel was increased to the maximum.
A new calibration technique was developed which nearly
eliminated the negative and positive problem
regions and concentric rings.
Systematic errors dropped from 20-30npB
to about
3npB
, where
one npB
unit is equivalent to a polarization brightness
of 
the brightness of the
solar disk. New calibration greatly reduced the radial rays.
Scans are now corrected to the average azimuth of CW and
CCW rotations which allows summing without
loss of spatial resolution. Summing ten or more images additionally
reduced the rays and brought the
random noise level down to less than 1npB
from its
original 3npB
level.
A technique which removes residual sky polarization was applied to each scan.
The 
centering system and guider cells were adjusted
throughout the year so that the solar
image and scan, respectively, stayed centered on the occulting disk.
Spikes in the intensity at some azimuths were discovered and
traced to unmasked holes in
the telescope.
Access covers were replaced to seal holes in the telescope.
A program was established to monitor the scattered light level
due to the 
and inform
the observers when cleaning was necessary.
This along with routine cleaning of the 
lens
has improved the calibrations.
Intensity measurements through the calibration tilt plates
revealed an unexpectedly
low transmission; the calibration tilt plates were made
of ``float'' glass which has absorption
bands in the passband of the instrument.
The index of refraction of these plates was unknown, making
the calibration factor derived from that
index similarly unknown.
Calibration tilt plates were replaced with
ones of a known material, BK7.