Hi Berk,
Indeed, when I say that the overlap is below 0.4 I mean the highest
values of the matrix of inner products. I expected the vectors to be
quite similar with their neighbours, i.e. I expected vector 5 to have a
high inner product with 5, or 4 or 6 of the other set. But this diagonal
trend, although present, is very weak, and is in fact more present for
the lower pc that the higher!
A part from the cosine content, we have checked the projections  of the
trajectory, and except for the 1st and 2nd pc's, the other exhaustively
oscillate around the mean value, which I though indicated good sampling.
Otherwise, how can I asses if the PCA makes any sense? (there are papers
that do PCA with shorter trajectories...)
Thanks!
Ramon