One additional effect which may need to be considered seriously in the future if a very narrow band pass approach is to be used is the Anisotropy of Anomalous Scattering (AAS) from a heavy atom in a protein. AAS occurs as a result of anisotropy of the local environment around the scatterer. The effects of AAS on the structure factor equation have been analysed in two articles by Fanchon and Hendrickson [32] and Kirfel et al [67].
The effect of AAS on the structure of the anomalous scattering factors in the region of absorption
edges can be very significant. One example is at the 
of Pt in crystals of
where the 
ions have square planar coordination. Anomalous scattering
factor have been measured [106] with the incident beam polarisation vector parallel and
perpendicular to the four-fold axis of the ion. The results indicated that the
inflection point of the edge shifts in energy by as much as 
between the two
orientations. The maximum value of 
at the absorption edge was also seen to change
by about 
. AAS effects have also been observed in the selenobiotinyl complex with core
streptavidin [52]. Shifts in the observed position of the Se 
edge energy of 
were observed when fluorescence was measured with the X-ray's electric field vector parallel to the
crystallographic 
and 
axes. The local environment of the selenium atoms in this case is
highly anisotropic due to the 
bonding.
Such marked AAS adds complications to the MAD method. Anomalously scattering heavy atoms with highly anisotropic environments which sit with different orientations within a crystal will tend to contribute to the diffraction pattern with difference anomalous scattering factors at particular energies in the region of their absorption edges. However the observed fluorescence spectrum from such a crystal sample would appear as the average of the two individual spectra due to the differently oriented heavy atom sites. This makes optimised energy selection more difficult.
The measurement of the anomalous scattering factors for a heavy atom bound in a protein are made even more difficult in the case of heavy atom derivatives prepared by soaking. In such cases the level of binding is generally weak and is maintained by an equilibrium between heavy atom in the protein solvent and bound heavy atoms. The presence of heavy atoms in the solvent is inevitable and will therefore contribute to the fluorescence measured from the protein crystal. The local environments of the heavy atoms bound to the protein and those in the solvent will differ in general, meaning that the observed fluorescence will not correspond to that expected from the bound heavy atoms alone. On the other hand, in cases such as in seleno-methionine containing proteins and metallo-proteins, the observed fluorescence should correspond to the bound heavy atoms alone since they are an intrinsic part of the protein and are not present in solvent. Also in the case of Hg binding to cysteinyl S, for example, the covalent binding should be strong enough to allow the Hg containing solvent to be replaced with a Hg free solvent without reducing the number of bound Hg atoms.
Anisotropic effects may be avoided by the use of suitable heavy atom complex ions, i.e. those which
are tetrahedrally or octahedrally coordinated to their ligands; one such example is the
ion used to prepare the lysozyme derivative used in this work. If isotropic metal
ion complexes can be used the problem of ions bound to the protein and ions in the solvent having
different fluorescence spectra is minimised. However it should be noted that there is a possibility
that the form of the fluorescence spectrum changes on introduction to a protein crystal. Some
evidence for this has been observed during this work where the maximum measured
value at the white line of the iridium edge in 
changed from 
e
to 
e on introduction to the protein environment.