Mike,

Different cameras were taken from different places under different lighting
conditions how can you say they are the same? For a comparison to be able to
be made that actually means anything all cameras would need to be tested
under the same conditions where as many variables as possible were
eliminated and that would include trying to use the exact same lenses. I
would expect the XTi to be better but he used a known very sharp lens to do
the tests on the XTi and a known average performer on the K10D, also the
shots for the XTi were very close shots where the K10D and D200 for example
were long range shots, how do you conclude that they were the same??

It is still very subjective, though, and there are *NO* dark areas in the
images.

From what I gather from people who have tested the K10D more objectively,
it is the #1 DSLR for dynamic range at ISO 100, at the pixel level.  The
1DSmkII might be close at the image level (dividing noise by the linear,
one-dimensional pixel resolution).

It has no real optimization at higher ISOs, though, like Canons do. Most
cameras don't; they have a single at-photosite read noise, and multiply
it in amplification, and then add the analog noise of the ADC, so the
total read noise as measured in electrons only improves ever so slightly
at higher ISOs, as opposed to dramatically with Canons.  For most of
these cameras, the only benefit of higher ISOs is the brightness of the
review image; under-exposing at a lower ISO may provide almost the same
quality, without sacrificing highlight headroom.  Quantization is not a
significant issue with these analog noise levels.

The standard deviation of an ISO 100 blackframe from the K10D (according
to my source) is about 0.9 DN/ADU.  The Fuji SLRs, the Canon 1D* cameras,
start at about 1.3.  Non-1D Canons are at 1.65 to 2.1, and most Nikons
are at about 3.0 or so.  Some cameras clip the RAW data at black, and
some don't.  These values are normalized to those that don't, as the
noise is only clipped near black in the cameras that do clip at black,
and not at higher luminance levels in any camera (blackframe noise must
be multiplied by about 1.64 for black-clipped RAW images).

With the K10D black noise at only 0.9 DN/ADU there is probably a little
bit of quantization noise, so ISO 200 probably is a bit better than ISO
100 under-exposed (ignoring the stop of highlights), so it has more value
than just a brigher review image or default conversion or JPEG.

The probably reason that the extra DR is not noticed in many DR reviews
is that the images are rendered with normal transfer curves, leaving the
extra shadow ranges down in the dark, where you can't distinguish them.  
You really need to boost the shadows to appreciate the difference.  There
is also the issue of where the camera exposes the middle grey in the RAW
data; this can vary quite a bit from camera to camera, and converters are
not consistent in the way they utilize RAW highlights.