>It is either (a) C2221 untwinned or (b) P21 twinned, but I _think_ if it is >really (a) then twinned refinement will give you ~0.5 twin fractions. If it is >really 0.5 twinned then the L-test (and the E^2 moments) should be really >clear: if not clear then it’s probably not twinned
I guess in both scenarios the refinement is carried out in P21 with twin refinement? But Wolfram's original post said that C2221 refinement yielded terrible R values, whereas P21 refinement +/- twin refinement was more in the normal range. Rmeas was also consistent with high twin fractions--at 50%, it should be about the same as the lower-symmetry SG. What more is there to discuss? Also, it is hard to see how refinement in the lower symmetry spacegroup with twin refinement should lead to 50% twin fraction--the mathematics are just different (intensity-additive versus amplitude-vectorial combination). And I definitely agree that the twinning tests should be roughly the same as Refmac. Try looking at the twinning tests at the *end* of the Aimless log rather than the beginning, and/or try Xtriage. JPK
