Hi Pei, Another parameter that affects the curvature of the background is where you choose E0 and kmin for the spline. Sometimes if you choose a value that is far down the edge, like half the edge step height, then the background has to curve alot to get through the edge region of the spectrum. You can see if this is the case by making three copies of the data set in Athena and then choose rbkg values of 0.8 1.0 and 1.2 and compare the chi(K) spectra produced with a low k-weight value. The region of the spectra where all three chi(k) data are the same should be used in the Fourier transform as kmin. That should help reduce the dependence of the chi(k) spectra on the rbkg value. Another approach is to bring E0 up the edge so that the background doesn't have to curve so much, and then make the same comparison. Good luck, Shelly ________________________________ From: ifeffit-bounces@millenia.cars.aps.anl.gov on behalf of ?? ? Sent: Sat 3/15/2008 12:34 AM To: ifeffit@millenia.cars.aps.anl.gov Subject: [Ifeffit] Rutile Rbkq remove question. To Dear authors: We have a Rbkg question about Rutile, and in order to demonstrate these questions more clearly, we make it up on attach .ppt file, and thanks for your great helping before. Pei ___________________________________________________________________ ?????????Yahoo!??????2.0????,???????! - ???? http://tw.rd.yahoo.com/referurl/mail/mail20/tag_hot0103/*http://tw.mg0.mail.... !