What do you do when XRD & EDS indicate different minerals?

I myself am very tired of "pyrochlores" (and maybe also a little of Lima da Faria) - as hinted by Pavel a pyrochlore formula you get from equalling the chosen elements (Ti+Ta+Nb+Fe3+) to 2. Now take almost ANY analysis made from ANY odd metamict ree-nb-ti-ta-oxide sample and treat it that way and VOILA - you have a "potentially new" pyrochlore. Thus, most analytical data from these metamict oxides will appear as pyro-group chemistry as the A-site and the anions are highly variable.

cheers

Uwe, I recently came across more information in response to your questions: "On how many spot analyses is the EDS-derived formula based? Was any chemical inhomogeneity or zonation observed?"

I read in another thread (Bogus formula accounting), I found a description of the method used by Excalibur (Tony Nikischer):
"In our EDS/SEM lab, when we run unpreped, single grain samples, we do so at least three times, with each grain either harvested from three areas of the original sample, or analyzed at three different areas of a single larger grain. This helps us evaluate homogeneity while rapidly pointing out any potential inconsistencies in the material and the data we have acquired."