Comment 16 for bug 1167036

Garth: no qualitative changes, "except" that (see my comment to Mikael) removing the Periodic specification from the .geo file now gives non-machine-precision errors, because the 'periodic DOF finder' now picks up the points that were previously too far apart!

Thanks for the advice regarding 'near'. near(x[0], 0) should be fine for my purposes, but near(x[0], 1e9) certainly wouldn't have been, and might explain some other issues I'd forgotten about.