Ah Stu - Trying to make me famous, are ye? I joined up to make a few notes I didn't happen to mention in a casual conversation last night.
The wee funnel works better than the feeder tank you see in the first pic. It's a slight airlocking issue because the piping in the panel wasn't 100% level. The pressure from the feeder tank didn't allow this to cure itself, and the thermosyphon effect just didn't start some days. The first sunny day with the funnel gave some spectacular geysers!
The Mark VIII is planned as a closed system with an expansion tank feeding in from the bottom end, but will need slightly better engineering (and a spare few quid for the tank). I'm considering using standard car anti-freeze as the fluid medium. I'd appreciate any thoughts on that idea.