I think Joel kind of answered this. It's to do with pivot-spindle distance. If the arrow is pointing exactly at the pivot point and either hole A or C is over the spindle then there is a right angled triangle with A (or C), the pivot and the intersection of the line A-C and the arrow line as it's corners. As A and C are equidistant from the arrow line the two triangles (A at spindle or C at spindle) will be the same - assuming the arrow is pointing at exactly the same point in both cases! It doesn't measure pivot-spindle as such but it includes it the process.

At least that's the way I understand it.