As for trying to work out where a fourth, missing stone was from the three remaining stones by fitting a circle to them doesn't work. There are loads of circle that will pass through the three stones, because you don't know which point to choose on each stone. Should the circle pass through the outermost point? The point directly below the centre of gravity? The inner edges? You will get a good idea of roughly where it could be, as gjrk says on this drawing - http://www.themodernantiquarian.com/post/55001, but it could have been a parallelogram. If it was then the fourth stone would have been off the circle somewhere between the top stone and the compass.
If you use the innermost edges of the three stones at Lettergorman then you get a totally dfferent circle to the one shown in the above mentioned drawing. Do that would make the plan trapezoidal.
Interesting stuff!