I suggest you go back to your images and see what is going on during indexing. You might want to check for certain unexplained peaks to could be an indication of reticular-non merohedral twinning: you have 30 58 118 90 90 90. If you take two unit cells and stick them on top of eachother, you will have a block with dimensions 60 58 118 90 90 90, which is allmost equal to a P422 symmetry. This could be difficult to handle in refinement, although not impossible in shelxl.
Other options could be that 1. Your space group is not P212121 but is P2x2y2z. Try all of them. The truncate NZ plot could be messed up due to data quality. 2. Your spacegroup is P2x11, P12x1 or P112x and you have twinning possibly combined with pseudo symmetry. Processyour data in all spacegroups or process it in xtraige and look at the R-values for various P21 choices. You might have to solve the structure in each and every space group and see if one of them does work out. HTH Peter 2007/6/8, Marius Schmidt <[EMAIL PROTECTED]>:
Dear Experts, I have a strange problem. Recently, we have collected some data on some myoglobin derivative. Crystals were quite twinned and we had to cut a leavlet out of a bundle of crystals. Crystal scattered to 2.6 A. Apparent space group P212121 with paramters [in A]: 30 58 and 118. One single mol-rep solution was found with sig to noise of rotation: 17 sig to noise of transl: 15 In two dimensions molecules pack nicely (see attached image vowi34.gif) in the other direction there is much to much space between the layers of molecules that one could envision proper crystal packing (see other image vowi32.gif). And indeed: refinement stops with R-cryst/R-free ~34/42 [%] with some spurious e-density outside the molecule which does not look like water. Matthew coefficient: 2.8 with one mol per asy unit. I tried to find another MR solution with the first one fixed and I failed (signal to noise of the best Rot and Trans solutions around 3). What could be the problem? I always thought that twinning cannot be the reason in P212121. Density distributions (from truncate) however seem to indicate some degree of twinning although this might also be a problem of data processing (data are quite weak). Maybe some of you have encountered similar problems and know how to solve them. Any hint very much wellcome. Marius