> Personally I find it disturbing to have the occupancy of "B 31" > set to 0.33 and that of "D 31" set to 1.00 simply because of an > insignificant shift in the position of the atom. > > Dale Tronrud
H2O on or near a 3-fold must be disordered because obviously H2O only has 2-fold symmetry. It could of course be H3O+ which does have 3-fold symmetry and which therefore could be ordered. We can therefore choose between 2 possibilities: either it's ordered _on_ the 3-fold with site occupancy=1 (site multiplicity=1/3), or it's disordered _off_ the 3-fold with occupancy=1/3 (site multiplicity=1). I completely agree with you that the deviation of 'B 31' from the special position is plainly not significant: certainly there is no way such a small shift could be detected (or indeed refined!), so we may as well assume that its symmetry does allow it to sit on the s.p.. So like 'D 31' it should be set exactly on the s.p. and its occupancy set to 1. Disorder should be indicated by fractional occupancy only if the molecular symmetry clearly demands it (and in this case it's not clear), or the disorder is observable in the map and it's possible to sensibly refine it as disordered. The difficulty comes if a refinement program refines an atom near the 3-fold exactly onto the 3-fold, and this atom either must be disordered because of molecular symmetry as in the H2O case, or the disorder is observable in the map. An atom which is just off a 3-fold would be likely to generate singularities in the matrix if you tried to refine its x & y co-ordinates anyway (you could of course fix them manually and not refine them, but then how do you decide what values to fix them at?). In that case it would be better to fix the atom exactly on the 3-fold and set the occupancy to 1, since then it can still be considered in reality to be the sum of 3 exactly superposed disordered atoms each with occupancy 1/3, which must however be treated in the refinement as a single atom with occupancy=1. Cheers -- Ian
