Several authors suggested that the thinning, with increasing depth of deposition, of calcite X-ray diffractometry (XRD) peaks obtained on planktonic foraminifera tests resulted from the preferential removal of their poorly crystallized parts as dissolution increases. By deconvolving XRD peak (104) from Globorotalia tumida (surface sediments, Sierra Leone Rise depth transect), we show that the full width at midheight does not depend only upon crystallinity, but reflects also the chemical and structural heterogeneity of foraminifera tests, which results in closely spaced, individual (104) diffraction peaks corresponding to phases with slightly different Mg contents. G. tumida contains two calcite phases: a well crystallized, Mg-poor calcite and a poorly crystallized, Mg-richer calcite. Increasing dissolution results in the preferential removal of the Mg-richer calcite and the improvement of its crystallinity, whereas the Mg-poor calcite does not seem to be affected by dissolution between 2750 m and 4950 m of water depth on Sierra Leone Rise.