The "Cimetidine-ExtractedIntensities" can be used with Fox
version 1.01 or above. To get Fox,
see http://objcryst.sourceforge.net OR on CCP14:
	 http://www.ccp14.ac.uk/ccp/web-mirrors/objcryst/
	 http://ccp14.sims.nrc.ca/ccp/web-mirrors/objcryst/
	 http://ccp14.semo.edu/ccp/web-mirrors/objcryst/

It does a Global optimisation using the extracted intensities 
from the powder pattern. 150 extracted intensities are used,
which is enough to find the correct model.
The cimetidine molecule was inputed using a ZScatterer (Z-Matrix)
description by using only information from the plane molecule
formula (ie bond distances, angles are given with tight limits,
and only some torsion angles are completely free. Look
in the .xml file to see all the limits).

[NOTE on Fox v<1.1, the SingleCrystal data is not displayed
in the Fox panels. But don't worry it's there in memory,
Display & the ability of adding yourself a (pseudo-)SingleCrystal
data object is available from Fox v1.1 and above].

First open the .xml file with Fox, then display the 3D structure to know
what you are looking for.
Randomize the configuration (Crystal menu:Parameters->Randomize Config.),
and you can launch the Global optimization.(Global Optimization
menu: Optimize->Run). The refinement is made on R, *not* Rw (works better).
Generally 10 million trials should be enough for a test (it takes around 
2 hours on a 1GHz PC with Fox 1.0.1).

If after 10 million tests you are still above R=.25, start again. It
probably means the N2-N4 branches still need to be exchanged and it
has not worked. With Fox 1.0.1 the correct configuration is found around
60% of the time. Further versions of Fox (1.1 and above) should go somewhat 
slower but with a higher success rate.

The best R you can obtain with the current configuration (no hydrogen,
guessed B-factors) is actually around 16%. Any configuration below
R=.20 should be good enough for refinement.