When I was doing the black dot analysis, I often had problem fitting the data. At one point I wanted to fit circles to the data, but that turned out too tricky. In conversations with Craig I claimed that some/many fibers do not really follow circles, especially during phi move. Craig pointed out that many fibers might look straight because they cover only a small angular portion of the circle.
I said I will look if I can find long/unusal cobra moves in the data which might indicate problems.