Hi Alejandro, A couple of things jump out at me: First, you say the data goes out to 10Ang^-1. That seems short for Cr in a Cr-Fe alloy, and implies either non-ideal data collection (as Sven alluded to: pinholes in a transmission sample, data collected in fluorescence, or a surface-contaminated e-yield measurement), or a highly disordered Cr environment.
4) About the nature of the alloy, from our experience we know that Cr is behaving in a very similar way to Fe here, so we should have a random substitutional alloy. From high resolution x-ray diffraction data we know that part of the Fe has not been alloyed, so there remain some 'clusters' of Fe.
To paraphrase, the sample is not a random mixture of Cr and Fe in a
bcc lattice, which seems a little strange to me, as bcc Cr and Fe have
very similar lattice constants -- I don't know whether they form a
uniform solid solution, but you are studying this sample, so it might
not be as simple as a random substitution, right??? Do you know
the fraction of Fe atoms that has been alloyed? Do you know how
much of the sample is crystalline, and what the defect density is?
These all might be important: You're assuming that all Cr is
surrounded by Cr, in a single bcc environment. Suppose there were two
or more Cr environments, with slightly different Cr-metal distances.
By assuming 1 environment, you'd end up with a sigma2 that included
the static disorder of having multiple environments.
.. which could all be related to having such a short data range.
By the way, any such issue would completely negate the assertion that
there is a simple relation between the diffraction