5/16/12
User:
I'm collecting some dataset of Mercury-soaked crystal and would like
    to have your advice because I never used mercury to solve the
    structure.

    From the fluorescence scan, I choose 12295 eV as a peak (Hg L-III scan
    is not like Se as you know) and low remote at 12200. I'm curious that
    people often use inflection for Hg-derivatizing dataset. Inflection is
    not very clear in my scan, either (two minima in close eV region:
    12281.88 &  12290)

    I'm wondering if collection of peak, inflection, and remote is a good
    strategy in general.  

James:
Yes, the Hg edge is fairly featureless.  
But even in the case of Se I still recommend two wavelengths over three: remote and "edge".  
By "edge" I mean a wavelength that is a compromise between 
maximizing the fluorescence signal and 
maximizing the slope of the fluorescence signal.  
Generally, this is about halfway between the "peak" and "inflection" 
assigned by Chooch (the program behind the Scan tab in Blu-ICE).  
I suppose I could change the default behavior of the fluorescence scan interface to do just that, 
but some people like to know the "peak" if they want to do SAD.

The good news about Hg, however, is that both the f' and f" traces are broad, 
so it doesn't matter so much exactly where you put the photon energy.  
This is true for most L edges.  
So, just put the "edge" wavelength at the point
 where the "edge" starts to drop off and Hg data should be good.

-James