Dear Izuok,
1. There are different metrics you can look at in order to know on how much anomalous signal you can rely: CCano,
ΔFano/σΔF,
and
ΔFano/F, In the CCP4 study weekend in Nottingham last weekend, Prof. Janet Smith said that for them it was the last one, ΔFano/F, the one working better as an overall estimate of the anomalous signal. Otherwise, you can look at the XDS output and see where the CCano drops below 0.3, or where the SigAno drops below 1... that should give you an idea. Otherwise put your mtz through Phenix Xtriage, it will output a good analysis of the data as well as reccommend a resolution cutoff for anomalous resolution.
2. I don't really know... well if you are working with a protein, there might be sites (Cys residues) where Hg would be more likely to be found... But I am not sure here... I would use as a first guess same number of sites as of Cys residues... but I am not working with proteins so I'm just guessing... probably there is a better way.
All the best,
Almudena