A critical survey of the literature has revealed that recent structure determinations and optical data are available for only three calcite-type carbonates: calcite, rhodochrosite and magnesite. For these structures, the limitations of the pointdipole approximation are discussed.The point-dipole model was introduced by Bragg (1924) to explain quantitatively the birefringence of calcite and aragonite with respect to the crystal structures. In this approach optical anisotropy arises from differences in dipole-dipole coupling along different crystallographic directions. This source of anisotropy is connected with the Lorentz-factor tensor. In addition, however, optical anisotropy may arise from an intrinsic anisotropy in the constituent ions. In this case electronic polarizability tensors should be considered.In order to investigate the relation between atomic arrangement and double refraction, the crystals of carbonates have been used repeatedly. The latest studies are those given by Isherwood &James (1976) andPohl (1978). In his paper Pohl (1978) has developed a method to determine the electronic polarizabilities of ions from structural and optical data by a least-squares fit. The method is based on Bragg's (1924) model of point dipoles and on an exact calculation of Lorentz-factor tensors. This model satisfactorily accounts for the double refraction of aragonite-type carbonates. In the present paper we discuss the application of the method to calcite-type carbonates.Structural parameters of only three calcite-type carbonates have been determined: calcite (Chessin, Hamilton & Post, 1965), rhodochrosite (Brown & Forsyth, 1967) and magnesite (Oh, Morikawa, Iwai & Aoki, 1973). Optical constants are given by Winchell & Winchell (1964). Application of Pohl's (1978) method yields the electronic polariz- Bragg's (1924) ideas, the birefringence arises from the oxygens within the isolated COl-groups: the stronger the birefringence the stronger the influence of the O z-ions on each other. Strengthening of the oxygen interaction can take place in two ways. Firstly, the distance between the carbon and oxygen can be diminished. Secondly, the polarizability of the oxygen ion can be increased. Since the C-O distance is fixed according to the structure determination, only an increased polarizability of the 0 2-ion remains. However, this results in an increased mean refractivity which in return is compensated by a smaller polarizability of the cation, thus explaining the above results.To overcome the shortcomings encountered with the Bragg model of point dipoles, two improvements may be considered. Firstly, the point dipoles can be positioned outside the atomic centers, because the outer electrons responsible for the optical effects may have a center of gravity distinct from that determined by X-rays. Secondly, an intrinsic anisotropy can be allocated to the polarizability of the ions. Both of these improvements involve the introcluction of at least one extra parameter. Thus the number of adjustable parameters will exce...