Thanks bemfarmer! I found it (by eye, without the script, that is too complicated for me ;-)
but it is funny, the outer curve is completely symmetrical (done with mirror) so I would expect that the error in the derived curve would be symmetrical too, and in that case the object should show as 2 curves, it should be broken at 2 places. that is why I did not think of an open curve at first.

thanks again!

wim

Tried a different offset. Seems like the intersecting line segments which are very nearly a straight line, but not quite, are
not being trimmed by the offset command ??

It's often times difficult for intersection mechanisms to get a clean intersection on things that are very nearly tangent with one another but not quite.

That's because things like that have a kind of "zone of intersection" where the 2 objects are within the target tolerance of each other across some extended area rather than at a sharply defined point.

- Michael