Dear Wim,
I suppose that the solvent content would influence the Wilson plot and
also the distribution of atoms in the unit cell, I would not be too
alarmed about the discrepancy.
Interestingly, the Wilson plot of the Fcalc values is about 60 A2 as for
Fobs in the output dataset.
As far as I understand, a refinement program alters the model so that the
calculated structure factors match the observed ones (this view might be a
little simplified for programs applying maximum likelihood), so your
observation confirms that the refinement program did a good job and is not
very surprising.
Tim