Yeah, that problem of a larger number not working all at once seems to be triggered when many of the non-intersecting pieces happen to be processed with each other first in the calculation, before getting to pieces that actually cut through one another.

I have a good example of it and I have spent some time working on it before but have not yet tracked down exactly what goes wrong.

When you do it in smaller batches it tends to be more likely to combine intersecting parts earlier in the process which seems to avoid the problem.

This kind of thing tends to happen particularly with union - with difference because you pick the objects in 2 batches with a base object and then other ones that are all intersecting it, that does not happen to run across the problem.

- Michael