Just to clarify... I believe that John and Gary's calculations are correct for the increase in spatial resolution of 10k vs 8k encoders. That is, if the scope is already pointing at an object, how much more 'accurate' is the coordinate reading with 10k vs 8k encoders? This is where you get the squaring effect because you're measuring the difference in the area of the "sky pixels".
My post was referring to the increase in pointing accuracy - i.e. you tell a GOTO system to slew to an object, by how much does it miss its target? Intuitively, you don't get the squaring effect in this case because each additional axis gives you new ways of missing the target: with 1 axis -> you can miss the target to the left or right; but with 2 axes -> you can miss it left / right / up / down... which cancels out the squaring effect.
Anyway, Peter, sorry for taking your thread off topic... I'll shut up now
