Al,

>My gut reaction says the first one (top left).

I'm thinking that'll be how most people perceive each slice to be 50% divided. It's just curious that we base that on the half of the radius. But I think people can visually measure ticks on a line (think ruler) much easier than they can visually measure area at least of a pie chart slice. That is I don't think most people can integrate the area under the curve, that's a much harder task than dividing a line in two.

>As an engineer, I know the second one is actually correct. I bet you've got a client absolutely insisting you do it the first way - I feel your pain :-/ It's an interesting artifact of human perception.

There's right, and then there's perceived right. *lol*

>Maybe you could try an offsetting visual trick. For example, light colours look bigger than dark ones - that's why all boats have white hulls. Maybe you could repeat the test with the inner slice black, and the outer slice white.

Actually these slices, there will be 1 to 8 of them, will be subdivided into 3 bands for early, on-time and late.

But what I really want is the instantaneous perception of the "timeliness" of the data with late being bad.

Another way to draw this would be with stacked bars, where one dimension is 0-100% for all bars divided into early/on-time/late, and the other dimension varies based on what would have been the pie slice percentages. In this case now you'd have a square divided up into somwhere between 3 and 24 cells.

Extend the original code with this and rerun it:

_screen.Box( 50, 500, 250, 700 )
_screen.Line( 50, 567, 250, 567 )
_screen.Line( 150, 500, 150, 567 )
_screen.Line( 150, 567, 150, 700 )

Here the vertical dimension represents the pie slice %, and the horizontal dimension the banding of the slice. Looking at the square for a while now that might be the more intuitive way to graph the data.