The problem also occurred in a program that would take fractional coordinates, so that's not the only snag.
I suggested another one:
From: Jacco van de Streek - The matrices representing the space-group operators are no longer diagonal, and the distinction between R*h and h*R (where R is the rotation matrix and h is the reciprocal space vector [h,k,l]) becomes important when calculating systematic absences.
And as a third possible source of error, I have noticed that trigonal space groups lead to a rounding error in the translational part that is different from other space groups. Which would make sense if e.g. the program internally defines the translational components of the symmetry operators as 0.3333 instead of 1.0/3.0 (not: 1/3).
Enough scope for errors. But I'm quite confident I finally got it right in Mercury 1.3, and Mercury is free.
Best wishes, -- Jacco van de Streek Research Scientist Cambridge Crystallographic Data Centre Cambridge, United Kingdom