1g1c was selected to be "lysozyme's evil twin" because the unit cell dimensions
of this structure almost exactly match those of tetragonal HEWL, except, of course
that the space group of 1g1c is actually P212121.  The real crystals were not
twinned, but if they were, the apparent space group would have been P43212.

To demonstrate this, a simulation of diffraction images was performed with
MLFSOM using the observed structure factors of 1g1c along with the anomalous
differences expected from Se had all the Met residues been SeMet.  Diffraction
patterns of untwinned intensities were computed with the crystal in two
different orientations.  These are eviltwinA_1_###.img and eviltwinB_1_###.img
Again, these two data sets are not twinned.  However, you may notice that
they look remarkably similar, this is because the difference in the crystal 
orientations is an exact 90 degree rotation about the 37 A cell edge.

By simply adding the images together from eviltwinA and eviltwinB, you arrive
at the diffraction pattern of a perfectly twinned crystal.  An example of this
is the eviltwin_1_###.img dataset.  The results of Elves processing (megrged
in P212121 for your convenience) are in eviltwin_5050.mtz.

  Other twin fractions can be had by simply changing the "mixing ratio" of 
the two images.  You can use the img_mix.com utility to do this. 
This was done to generate
eviltwin_8020.mtz and eviltwin_6040.mtz

Regardless of the twin fraction, you should have no trouble finding a molecular
replacement solution, and subsequent refinement in TWIN mode should work.

The difficulty comes with trying to solve the structure using the anomalous
differences.  Despite the relatively strong anomalous signal here, a 50:50 
twin ratio seems to be impossible to solve.  60:40 is just barely do-able.
Can you get 61:39 to work?  Try it and see.