But that's NOT what's being illustrated here because the samples have wildly different levels of fuel in them, most likely due to varying operating conditions during the periods in which they were run.
The other issue of course is the visc accuracy in these tests. If we see significant deviations from the blender's spec sheets, that should be cause for concern as well.
Basically, there's a huge stack of variables here that are not being accounted for, certain data being soundly ignored, and yet "firm" conclusions are being drawn, which is utterly bizarre.
Let's, for a moment, assume that the numbers from Blackstone are all spot-on for viscosity:
1. SuperTech went from 11.1cSt to 9.62cSt. With a flashpoint of 440F, there's no fuel in that sample, so the 13.3% viscosity loss is all from mechanical shear
2. Pennzoil U went from 9.48cSt to 8.87cSt. With a flashpoint of 395F, there's absolutely fuel in that sample, so the 6.4% viscosity loss is at least in part, from dilution
So, despite the fuel dilution, the Pennzoil product experienced 50% less viscosity loss, percentage-wise, than the SuperTech, assuming again that the Blackstone numbers are accurate.