Chances, You should read this paper by Shelly Kelly: http://dx.doi.org/10.1016/S0016-7037(02)00947-X It shows an example that is, I think, similar to your question. She fits an unknown by combining paths from different FEFF calculations. B On June 29, 2014, at 2:19PM, ZHAN Fei wrote: Dear all, I have a confusion.For instance,when I want to fit a simple which has Pt-O,Pt-S,Pt-N,Pt-Pt cordination shell;when can not find a compound with all same shell or can not find one with all suitable bond length,can I use different scattering path from different model to fit diffferent shell? and if do so atom may in different pot in feff ,are the phase and amplitude still transformable? Other choice is to use path-like object in demeter to construcr a more suitable distance scattering path,or construct a new cluster model according to several cif files, myself,which proticol is feasible or better ? sincerely zhanfei