The crystal structure of a magnesian cronstedtite-2H2 from Pribram, Czechoslovakia, was refined in space group C1 to a residual of 5.4% with 1832 independent reflections. Tetrahedral ordering between Fe3+ and Si is judged to be complete on the basis of electron density maps, the first confirmation of such ordering in a layer silicate. Octahedral cations are disordered on the M sites. Mean T-O bond lengths do not confirm the tetrahedral ordering, perhaps due to tetrahedral distortion in order to relieve the extreme corrugation of the sheet that would arise from ordering of such different size cations.

In the initial stages of refinement the structure of the crystal under study could not be described adequately on the basis of a single 2H2 polytype. The structure contains two kinds of 2H2 domains that appear shifted by b/3 relative to one another due to a mistake in the interlayer stacking sequence. This mistake of zero shift (the normal 2H2 polytype consists of alternating —b/3 and +b/3 shifts) affects only the tetrahedral sheets. It creates adjacent enantiomorphic domains of 2H2 packets such that domains with small Si tetrahedra sit above larger Fe3+-rich tetrahedra and vice versa. This mechanism to relieve strain may indeed be the cause of the stacking mistakes. A third kind of domain contains regions in which the sense of tetrahedral rotation is reversed, giving rise to split basal oxygens on electron density maps. This multiple domain model best explains extra atomic positions observed on Fourier maps, lack of streaking of k ≠ 3n reflections, and possible nonintegral “satellitic” reflections observed on Weissenberg films.