Strong fractionation of deuterium in photolysis of
H 2 O and above the hygropause reduces the production of HD relative to H 2 on Mars by a factor of 3.7 total. The model by Y. L. Yung et al. (1988, Icarus 76, 146-159) for deuterium fractionation in chemical reactions on Mars corrected for this factor results in (HD/H 2 )/(HDO/H 2 O) = 0.43. This value may fit the deuterium abundance observed by V. A. Krasnopolsky et al. (1998, Science 280, 1576-1580) if the eddy diffusion coefficient does not depend on solar activity: K = 1.4 Γ 10 13 n -1/2 cm 2 s -1 (model 2). The Mariner 9 observations show very low variability of atomic oxygen at the 1.2 n bar pressure level (h βΌ 125 km) with solar activity. This requires eddy diffusion to be proportional to the solar activity index F 10.7 cm : K = (F 10.7 cm /30) Γ 10 13 n -1/2 cm 2 s -1 (model 1). The fractionation factor for escape of hydrogen isotopes is equal to 0.016 and 0.135 for models 1 and 2. These values have been averaged over the solar cycle. The threereservoir model for hydrogen isotope fractionation suggested by Krasnopolsky et al. (1998) involves a reservoir composed primarily of water ice in the polar caps that isotopically interacts with the atmosphere. Assuming that water ice is half of the total volume of the polar caps and the polar-layered deposits, the total loss of water from Mars is equal to 65 and 120 m for models 1 and 2, respectively. Along with thermal and nonthermal escape, these values may include the loss of water by oxidation of regolith, if the released hydrogen escaped with isotopic fractionation. Although the solarwind Ξ± particles are the main source of He on Mars, capture of the solar-wind H + and D + ions by Mars has a negligible effect on the thermospheric abundances of H and D. Improved observations of minor components in Mars' thermosphere may resolve the problem of eddy diffusion at various solar activity and choosing between the models.