standard deviations for cell constants

Zbyszek Otwinowski zbyszek at
Thu Jun 6 09:57:40 EST 1996

Tricia Takahara wrote:
> Has anyone ever reported or determined a standard deviation
> for unit cell constants determined by using an image plate?
> We have a Mar and use DENZO to process the data and would like
> to report some sort of standard deviation for our cell constants
> in our next paper.  The problem is that we use a capillary mount
> and move the plate in and out when the crystal starts to decay
> and we have to translate it and re-center it.  This means we
> have to refine the distance several times during the DENZO
> processing.  This results in slightly different cell constants
> for different sections of the data (the constants vary by a
> tenth of an angstrom or so).  It has been suggested that we
> just average the cell constants and do a standard deviation
> calculation by hand, but we are not sure this is correct because
> we vary so many parameters during the processing of the frames.

Companion program to Denzo - Scalepack can do global refinement
of unit cell parameters. Global refinement in Scalepack uses
only intensity information, so it does not depend on detector
parameters (e.g. distance). Global refinement (postrefinement)
is very precise. Scalepack (optionally) can estimate standard
deviation of unit cell parameters. These estimates, unfortunatelly
are sometimes too low due to systematic errors. The source of
systematic error in postrefinement is goniostat. In practise,
many goniostats/shutters are far from perfect and errors in
goniostat are hard to predict. If crystal is not frozen, additional
two sources of error should be considered:
1) Unit cell changing due to radiation damage - rare, but sometimes very
2) Crystal slippage - crystal slippage can correlate strongly with
unit cell parameters in low symmetry space groups.

For example, in orthorhombic space group, for 2.A data, 90 degrees
of data, sigma of unit cell should be less than 0.01A.
Systematic errors tend to be larger. Unfortunatelly, it is very hard
to estimate magnitude of systematic errors.

There is no sensible procedure for reporting
error of the unit cell. Fortunatelly it does not matter, as the error
is very small.

  Zbyszek Otwinowski                     |        zbyszek at
  University of Texas                    |        tel : (214)-648-5098 
  Southwestern Medical Center            |        fax : (214)-648-5095 
  5323 Harry Hines Boulevard  Dallas, Texas  75235-9038

More information about the Xtal-log mailing list