Kyung-Mun Byun,

When performing a multi-peak fit using Analysis:Fit Multi-peaks:Gaussian, the nonlinear curve fitter runs in the background. Since this is the case, you have manual control over whether a parameter is allowed to vary or remain fixed. To fix the two peak positions at 940 and 985, go through the following steps:

1. Perform the fit as you normally would by selecting Analysis:Fit Multi-peaks:Gaussian and going through all the intermediate dialog boxes.
2. Return to the graph window once the fit has taken place and select Analysis:Nonlinear Curve Fit. This should automatically bring you to the Fitting Session dialog box.
3. Scroll down through the list of parameters and enter 940 and 985 into the xc1 and xc2 Value text boxes. While doing so, deselect (uncheck) the Vary? check box next to xc1 and xc2. This fixes the two center positions.
4. Click the 10 Iter. button to perform additional iterations using your center positions.
5. When you are satisfied with the fit, click the Done button.

I hope that this information proves helpful.

Sincerely,
Ryan Toomey

quote:

---

Statistical tests with error values from gaussian fits.

I have no problem with the multiple fit in 7.0 (works better than in 6.1). However, can you explain what the error values reflect for the three parameters in each Gaussian (mean, amplitude, width). Thus, in a recent gaussian fit I may get xc1 = 13.6 +/- 0.2. Is the +/- value the standard error of the peak estimate? If so, can I use it to test if this mean is significantly different from a similar fit to a second data set. If it can be used then what are the degrees of freedom for the lookup tables and how are these affected if I do multiple gaussians on a single data set?

Thanks
Elise

---