It was considered that experimental conditions preventing any external gas diffusion [7] and the systematic use of non porous samples should permit the study of the chemical process. It is assumed that the rate of reaction (linear rate of corrosion: v1c) is uniform on all the surface of the crystals, and remains constant during the process. Besides, the diamond particles are considered to be cubes of edge a, for which the rate of movement of the reaction interface is va. This is indicated in Fig. 1 and in eqn (1) expressing the roughness factor r which relates a sample to the corresponding model. The results presented in Table 1 show that this roughness factor does not change with the granulation or with the rate of movement or “wearing” U: the gasification then takes place only at the geometrical surface of the diamond particles. The cubic model and the considered hypothesis allow the a priori calculation of eqn (2), representing the variation of the rate V (expressed in % of the initial mass having reacted in one hour as a function of time). Figure 2 shows this evolution with the advancement of reaction U. When the reaction has been initiated (U > 10%), the theoretical curve seems to have the same shape as the experimental curves. By changing coordinates, eqn (2) leads to eqn (3), which should be graphically represented by straight lines. These are indeed experimentally observed (Fig. 3). The introduction of the parameter a0 in the coordinates implies that the transformed curves corresponding to different particle sizes are identical. In Fig. 4, it is shown that the fit is quite satisfactory considering the error associated with the determination of a0. The initial value of the linear rate of corrosion v1c (eqn 4) and the mean value of the rate of movement of the reaction interface va (eqn 5) are obtained by extrapolation of the linear transformed curve. In good agreement with the hypothesis va = r · v1c (Fig. 5) the equation of the transformed curves may then be simplified (eqn 6). The straight lines form a bunch and their geometrical envelope may be calculated (eqn 7). This is important because of the respective positions of the transformed curves and that of the cubic function: this is an experimental verification of the adopted model. It can be seen in Fig. 6 that the transformed curves and are tangential, which indicates that the model may be applied to the gasification of natural diamond by O2 and CO2. On the contrary, the oxidation of graphite, prepared from diamond, leads to the transformed curve : it is obvious that the model cannot be applied. It may be noted (Fig. 7) that, in this case, the transformed curve is not absolutely linear. The comparison between the oxidation of diamond and the oxidation of graphite indicates two important differences: 1. (a) diamond is much more reactive than graphite, as shown by the values of the linear rate of corrosion presented in Table 2 (the values for the three graphites were calculated from the data of Bonnetain and Hoynant[2]. This difference in reactivity might reflect the structure difference [3]. 2. (b) In the case of diamond, the apparent order of reaction associated with the oxygen pressure is low, about 0.25, while 0.5 is the value generally considered for graphite. This may be compared with the oxygen absorption measurements on diamond, carried out by Bansal et coll. [1] and Sappok and Boehm[11]. A tentative confrontation of our results and those of Evans and Phaals[5] is presented in Table 3, as far as the experimental conditions are comparable. In conclusion, it should be stressed that in spite of the good agreement of the experimental results with the adopted model, scanning electron microscope examination of partly oxidized samples shows selective attacks in some points of the crystal surface, these being an evidence of the variation of reactivity with the orientation of the crystallographic planes [5].