Crystallite size is the mean extent of coherently diffracting domains within a powder or polycrystal, a quantity distinct from particle size because one particle may contain many such domains. Diffraction line broadening is the usual access route: the Scherrer relation converts peak width into a domain dimension, but the measured width also contains instrumental broadening and depends on crystallite shape through the Scherrer constant. Instrumental broadening should therefore be corrected using a suitable standard, LaB₆ among them, or an appropriate profile model before a meaningful size is extracted. Extensions of the Scherrer analysis can, in favorable cases and with additional assumptions, yield a crystallite-size distribution rather than a single number; such distributions are not generally obtainable from peak widths alone and require suitable whole-profile or envelope-function treatments ^[1,2]. Size broadening overlaps with strain broadening in the same profile, so separating the two requires several orders of reflection or a full-profile treatment; established approaches include Williamson–Hall analysis and Warren–Averbach Fourier analysis ^[3]. Different routes to the same quantity give systematically different answers, as comparisons on hydroxyapatite show ^[4], and the same caution applies when the phase sits in a deformed matrix ^[5]. The resulting value is best regarded as an apparent domain size whose physical interpretation depends on how much of the profile has been modeled and on the corrections applied; it should not be equated with a unique physical dimension unless those effects are adequately treated. Machine-learning methods are also being explored for diffraction-data analysis and for assisting parameter extraction, but their predictions depend on training data and should be validated against physically based profile models rather than regarded as a substitute for instrumental corrections ^[6].