Interferences (overlaps) occurring when lines of other. elements affect eit
her peak or background measurements can cause errors in quantitative WD ana
lysis, bur may be minimised by suitable choices of analysis conditions such
as spectromerer crystal, background offsets, and pulse-height analyser set
tings. Computer spectrum-simulation is much more effective than reference t
o wavelength tables for investigating interferences. The 'Virtual WDS' simu
lation program developed by the present authors, hitherto applied only to '
ordinary' elements (Z greater than or equal to 11), has been extended to li
ght elements for which evaporated multilayers are used in place of true cry
stals. 'Virtual WDS' utilises experimentally recorded light-element K spect
ra and L and M spectra of heavier elements in the same wavelength range. It
is impractical to record all high-order peaks, so computed line profiles a
re used, with widths and intensities interpolated from a limited set of obs
ervations. The relative positions of first and higher order peaks are affec
ted significantly by the refractive index of the multilayer, requiring modi
fication of the Bragg equation. Suppression of high orders by pulse-height
analysis is less effective than for 'normal' wavelengths, owing to the brea
dth of the pulse-height distribution for low X-ray energies. Simulation usi
ng a Gaussian expression aids optimisation of threshold and window-width se
ttings.