I am sure you have checked this but
A) is there non cryst translation? If so this can make space group
selection tricky . Could it be P3 21 or P31 21 or ...
B ) the twin indicators at data processing are pretty reliable and if that
suggests no twinning it probably isn’t there... you could process
The twin fraction can vary during data collection - if the beam is smaller than
the crystal
For example you could have crystals which are like two very short ends of
pencils - with flat ends together (in your case these might pencils could be in
P3)
In some orientations you could have the beam g
Just double checking that Refmac automatically handles twinning without
the need for a keyword or a box click in i2. I have a crystal in P32 21
that gives poor Rfactors and maps (Rfree mid 40s) from what should be an
excellent model. I have reprocessed in P32, P1 and C2 but all give
similar re