This paper shows how signal-to-noise ratio, shift, shape and contributions from the factors present in a set of spectra are related to one another from the point of view of the ability of principal component analysis (PCA) to detect the presence of the factors. In the derived relationship a coefficient with a value dependent on the number of parameters appears. Values of this coefficient for a particular criterion for detection of the factors and for four selected factor shapes have been found from the results obtained for simulated sets of spectra of known properties. Consequently, a simple formula for evaluation of the minimum shift between factors (d min ) that ensures the presence of both factors can be proposed. This paper also proves that d min can be lowered when the spectra are recorded in derivative form, and that further lowering of d min can be obtained when PCA is applied to the set of spectra after rejection of their less informative parts. As an illustration of the latter effect, d min values of 0.125 and 0.035 eV, respectively, are obtained for a set of derivative Auger MNN spectra.