Mathematical isolation of component spectra in HPLC/UV-vis and GC-MS. How unique are the resolved spectra?

The resolution of overlapping spectra in GC-MS and HPLC/UV-vis is fundamentally limited by the quality of the experimental data. The narrowness of the solution range depends on the degree of overlap between components. If the components are dissimilar, the solutions obtained by all mathematical methods are robust. Small perturbations in the observations do not change the calculated solution very much. Alternating regression (AR) is a useful tool in the analysis of overlapping spectra because AR can be calculated very rapidly. The robustness of the solution can be easily checked with AR. The mathematical analysis is repeated several times after adding different sets of noise. Each time different random spectra are used as a starting point. The range of solutions thus obtained reflects the quality of the data for resolution purposes.