A new, accurate, low-temperature X-ray analysis of lithium tetrafluoroberyllate, Li2BeF,, reveals that the charge density in this crystal is better represented as a superposition of spherical neutral-atom charge distributions than as a superposition of ionic charges. The two distributions are so similar that they are hardly distinguishable by examination of the total charge density in real space. However, the two models are clearly differentiated by analysis of the weak low-order reflections; the measured intensities are reproduced better by calculations based on neutral-atom scattering curves. This result may appear to run counter to the current conventional wisdom concerning ionic solids, but it is unlikely to have any important practical consequences.Although many solids, such as LiF, are conventionally described as ionic, an experimental proof that they are really built from ions is hard to come by. In fact, the ionic and covalent descriptions are not so very different. As Sluter [I] has pointed out, the difference in charge distribution between the superposition of the neutral atoms and the superposition of the ions is small and subtle, and difficult to determine by examination of the total density. The densities of individual neutral atoms and corresponding ions only become appreciably different at such large distances from the respective nuclei that in a condensed phase, where densities of neighbouring atoms overlap, this diffuse difference density cannot be unequivocally assigned to one atom or another. At these large distances, the density associated with a given atom may be quite small, but the contribution of the spherical shell to the integrated charge may be considerable.The possibility of solving the problem in reciprocal space would appear to be just as unpromising. 'Any attempt to determine the state of ionisation of the atoms in a crystal by means of measurement of the atomic scattering factor is likely to fail, since . . . the curves will differ appreciably only at (scattering) angles for which no spectra exist' [2]. This authoritative though pessimistic conclusion has been reiterated many times, most notably, perhaps, by Bijuoet and Lonsdale [3] in their discussion of earlier claims to have determined the state of ionization of LiH. The point is that the scattering powers of neutral atoms and their corresponding ions differ appreciably only in a small region of reciprocal space close to the origin.Nevertheless, the scattering factors of charged and uncharged atoms do differ appreciably at small scattering angles and are not quite identical even at larger ones. Thus, James's conclusion may not necessarily be valid, especially for crystals with unit cells large enough to produce diffracted spectra at small sin QjA values (< cu. 0.2 A-') and provided the structure amplitudes can be measured with very high accuracy. Over the years many attempts to overcome these problems have been made (e. g. [4]). Other aspects of the difficulties of assigning atomic ionicities in solids have been reviewed recent...