Chris,
Good to hear back. The tolerance figure determines how close to an actual intersection CB may decide (on its own) to break a line, based upon (I suppose) internal vector points or some-such similar points in the graphic primitives.
For instance, if you 'explode' your circle, you'll see it's made of three segments, and my presumption is that if the tolerance is set high enough, it might decide that the end-point of one of those segments is a 'suitable' place for an intersection -- even though we know it's not one.
If I do a break at intersections with ctl-B and the intersection tolerance set to 1.0, the circle segment end-point at -41.15, -1.991858, 0.0 becomes one of the 'cusps' of that bad break. It seems to work differently if I use the menu version of break at intersections (but still not correctly, until the value is set very small.)
To see it happen, draw a one-point point-list. Then edit the point so it resides at -41.15, -1.991858, 0.0. Set your intersection tolerance to 1.0, select your circle and line, and do a ctl-B. Cute, huh? Interestingly-enough, the OTHER segment end-point at -41.15, +1.991858, 0.0 is another cusp in that improper break.
It's not something the user can take direct control of, so it's always a good idea to use a small value for the intersection tolerance.
Lloyd