Measuring the areal density of solute atoms contained within a narrow planar fault, boundary or quantum well by standard analytical electron microscopy is possible but often yields a significant systematic underestimate. Improvements of detectors and/or aberration correction will improve resolution and counting statistics but not these systematic errors. Over the years, a new way to analyse series of analytical spectra by linear least-squares fitting approaches has been developed that eliminates this problem. The new method, presented by Walther on pp. 310–313, is more precise and reliable. The method has been shown to yield reliable segregation values down to a few percent of equivalent full monolayers, typically corresponding to <1 atom/nm², and the precision can be directly estimated from the R² parameter of the linear fit. It can be implemented in different illumination modes and is equally well suitable for X-ray and electron energy-loss spectroscopy. The figure displays an annular dark field image at the top left where three ultra-thin InAs quantum wells (with nominally 1.6, 1.8 and 2.0 atomic monolayers of indium) appear bright relative to the surrounding GaAs due to their enhanced scattering contrast. The coloured maps are intensity maps obtained for various X-ray lines, using a standard energy-dispersive Si:Li X-ray detector. From the linear fit of a plot of the As/In ratio as function of the width L of the sections around each quantum well, the total amount of indium in each quantum well could be determined to about ±0.1 monolayer precision.