Thanks adrianh++.

I should've spotted the shift thing myself. It was 4:00 am though. (Excuses, excuses &^) ... that's a crossed-eyed smiley).

The $self = bless \$self, $class; bit was just an idea that I never got around to looking at the effects of as the other stuff was going wrong. The strange thing is that using that rather than [], meant that somethings appeared to be working, ie. the hashes in the base class were being populated, though 2 of the 3 hashes were getting two entries. 1 under the subclass reference and one under the base class reference? The 3rd hash (%approved) was only getting a single entry.

I know that all sounds implausible given they are all initialised in the same way, but that is what I was seeing.

I've made a backup of that version and will get back to investingating what was going on once I've finished experimenting with this.