As you know, F5 and F6 uses a large internal database to match the current photo against it for proper exposure. The first photo up in the mountains is very suspicious. Could you have possibly found the one scenario the F6 doesn't have
in it's database (which usually amounts to one bad exposure in a thousand rolls)?

I would say these are all too far off for an F6. An F5 would have never
been that far off. Could exposure comp possibly have been bumped to -1?
Was there anything attached to the lens such as extension, lens shade, etc.?

If you are 100% sure that all settings were appropriate, here's what I would
do: Recreate a scene similar to the first photo using small props which would
cause the camera to not expose correctly. I know it sounds like a lot of work but
believe it or not it only takes a few minutes. This is what I do with all my
new cameras. Take a few photos and play with exposure comps and other adjustments
one at a time and write them down. See what changes occur. It would really be
helpful if you could borrow another camera like an F5 or an FM3A and a handheld
meter to compare.

Or...

Another much more accurate method to test the camera would be to acquire an
IT8 TARGET for test photos.

http://www.eximvaios.com/catalog/images/kodak_IT8.jpg

Also, I almost forgot the most obvious question. Where and how were these
developed? I usually shoot at least one or two photos of the IT8 target
at the beginning of every roll just to make sure the developer doesn't
screw it up.

A and I has worked out pretty well. Also Hamilton Color Labs here in
Omaha does a pretty decent job.

http://www.hamiltoncolorlab.com/

http://www.aandi.com/