I didn't check anything before I noticed the problems. By 'clean up' I was formalising place names and modifying user fact types so that they scanned better in the left hand panel (under the 'person' tab).
The only four errors reported, upon import, are still there and they are to do with christening dates:
29 Feb 1819, 'same day' and death 'in infancy'. The corrupted relationships are showing up in the main panel (in family listing) that shows the pedigree of the individual.
People workspace, person tab, relationship button: showing correct biological parents and siblings but then added all the children from a different family (with a different surname) as biological.
I'm consulting FTM side by side with PAF, so I know that the data in PAF was and is correct.
I imported the data as a paf file which was a straight copy:
Plan/New Tree/Import a tree from and existing file/(browsed to the PAF5 file)(gave the tree a new name)clicked continue. FTM analyses then imports. I'm doing it again so that I can describe to you exactly what happens!!
Not that it's likely that the same error would occur with the same individual twice - but I'll check in the new tree.
Well, that's interesting. Same individual showing the same wrong family!
Might it have something to do with the RIN and MRIN numbers from PAF? After compacting, it reports that the file size has been reduced by .47%. Still showing corrupted relationships.
I think I'll try starting a new tree with a gedcom exported from PAF.