I think that this is an important topic, because it strikes at the core methodology for measuring both gamma and color performance. Thus, I spent some time taking some more measurements. This time I used a more accurate instrument (Minolta LS-100) and I performed tests on both the Pioneer 5020 plasma and a Sony CRT.
First, let's begin with my initial measurements on the 5020 using the Chroma 5 colorimeter. I'll use the same % difference graph that you did.
Here's the same measurements taken with the Minolta.
The % difference curve now looks a little different. There's still the same raggedness at the bottom and the output is still higher at all levels of stimulus, but other than that there's no discernible trend that that I can make out.
Here's the gamma results from the same data.
Looking at the gamma curve, there appears to be a small effect. From 70% up as the gamma for the 1% pattern becomes progressively higher until at 90% it is 0.10 higher.
Here's the data for the Sony (uncalibrated) CRT.
There is clearly a trend here. The % difference between 1% and 15% patterns becomes progressively lower as the signal level increases. Here's the gamma chart for the same data.
Here there is a small difference, but it is the opposite of what we saw on the plasma. The gamma values are slightly LOWER and throughout the entire range.
Just to make clear to those who may not appreciate this, CRTs and plasmas are voltage limited in way that requires the use of window patterns. These displays simply cannot produce the expected output at higher levels of stim using full field patterns. For example, here's the Pioneer with full field patterns (window before, full field after.
And the gamma results for the same data.
The results are clear. The voltage limiting effect produces a LOWER gamma throughout and LESS output only a higher levels of stim.
What to make of this? I honestly don't know. The data are simply not consistent. The only consistent results I see are:
1) Measurements using 1% patterns yield slightly higher values at all levels of stimulus.
2) Whatever differences we do see are very small, and are not the result of voltage limitations of plasma and CRT displays.