The experiment described in the previous section was repeated for shuffle neighbourhoods with radii of 1 (Euclidean distance), 1.5, 2.1, and 3.7, to test the hypothesis that this would extend the range over which different degrees of mis-registration could be discriminated.