If the result is satisfactory, the spectrum is further analyzed by more sophisticated means, such as model-fitting by Gaussian curves. One defines the number of curves to be used (20); 0 means the automatic minimization with a single Gaussian, without initial values for the parameters. The GAUSS command (21) activates the minimization algorithm, which finally converges (yes it does !). One plots the fitted Gaussian by the FIT command (22). The residuals may also be determined by the RESIDUAL command (23) which first copies the R "memory" into the T "memory", then subtracts the Gaussian. These residuals are then plotted (in the current limits) by the PLOT command (24).