Ya know, I love discussions like this. By that sort of math, my gun would barely be able to hit the target, much less put 'em in the X-ring on anything resembling a consistent basis. gun)
Actually... errors don't stack linearly like that. In theory, they maybe *could*, but your chances are probably better of hitting the lottery. Root Sum of Squares (RSS) is a better method for figuring out how errors add up:
TotalError = sqrt((Error1)^2 + (Error2)^2 + (ErrorN)^2...)
or in this case...
sqrt((10)^2 + (6.5)^2) = ~11.9
Then add in that E.S. is a lousy measure of consistency, mathematically speaking. It's easy to calculate and simple for the average person to understand, but it gives too much (all) weight to two 'outlier' data points - the ones statistically least likely to occur on a regular, repeatable basis - so the actual effect on target is (in my experience) rarely as drastic as predicted by simply plugging the high and low ES numbers into a ballistics program.
There's something to be said for considering absolute worst-case scenarios where you would experience the actual total ES spread between two sequential shots, stacked squarely against the absolute worst possible spread between shots in the group... if you like worrying yourself into a lather and an upset stomach
YMMV,
Monte