The reported occurrence frequency of twinning in olivine is at odd with the degree of atomic restoration across the composition surface obtained by the analysis of the pseudo-eigensymmetry of the crystallographic orbits: the {0 3 1} twin should be the most frequent one, whereas it is reported as the rarer. The possibility of incorrect indexing, already raised in previous reports, is considered highly probable due to the fact the {0 1 1} and {0 3 1} twins have exactly the same twin lattice and differ for the plane acting as twin element. The rarity of {0 1 2} twinning is explained by the small degree of atomic restoration and the large separation between the planes that produce that restoration. A fourth twin law, reported as (1 0 0), is impossible unless the structure is merohedral and is probably due to errors in the indexing of crystal faces in the old morphological studies.