If the volatility vs flash point were plotted for PAO 4, 6, 8, and 10 cSt viscosities, I would expect a curve could be fitted through the points with all points close to the curve. If the same was done but using Group II, I would expect the curve to not fit through the points as well as they do with PAO, because of less consistency in molecule type and size in Group II. With motor oils containing a variety of base oil types and viscosities of each type as well as additive carrier oil and additives, I would expect a trend to still show in the viscosity vs flashpoint plot, but expect it to have considerably more outliers than I’d expect in the plots of purely base oils of same type.
One thing that will cause outliers is an oil with a small percentage of relatively high volatility. It will evaporate rapidly in the flash point test, decreasing flash point. In the NOACK test, it will evaporate out early in the test and the test keeps going without this volatile portion continuing to cause mass loss.
I would be interested in seeing someone make the plot with a bunch of different brands of motor oils.