Five color/albedo units, including polar frost, have been recognized and mapped in the southern layered deposits on Mars. Atmospheric dust scattering was measured in shadows and modeled in order to remove the component of brightness in Mars images due to the atmosphere and quantify the albedo and color of the surface. The layered deposits appear to be mantied by red dust, except where eolian stripping has exposed the underlying bedrock. Frost and bare ground are mixed below the resolution of the images in many areas adjacent to the polar cap, some of which appear to be younger than the surrounding layered terrain. Dark material has been deposited in topographic depressions in much of the south polar region, including the layered deposits. The available observational data suggest that the layered deposits are composed of bright dust, ice, and a small amount of dark material. If the dark material is sand, a periodic change in polar winds seems required in order to transport the sand poleward into the layered terrain. In any case, the observations are not consistent with the layered deposits being composed only of bright dust and ice. The Mars observer camera and infrared instruments should be particularly useful in this investigation. Viking Orbiter 2 imaged the south polar region in three colors, typically at a resolution of a few hundred meters per pixel. In this paper we first describe how these color data were processed and then discuss the effects of atmospheric scattering on the apparent surface brightnesses. We modeled the dust scattering in shadows and used the results to estimate the atmospheric component of brightness in the images (details of the method are presented in the Appendix). The dust-scattering properties found in our analysis are similar to those found in previous studies. Removal of the atmospheric effects from the color images then allows esti- The images were radiometrically calibrated in Flagstaff using the Planetary Image eartography System (PIES), so that the data value in each pixel represents the actual reflectance (l/F) observed. The relative calibration error measured where the images overlap is as large as 10%, so that the absolute uncertainty in calibration is no less than 10%. This result is consistent with the 13% (let) absolute uncertainty near midscale in the Viking television calibration reported by Klaasen et al. [1977]. No estimates of the radiometric accuracy of PIeS have been published, and the absolute uncertainty of the Viking Orbiter calibration may be greater than 13%. Preliminary analysis of Viking Orbiter 1 Phobos images shows differences of up to 18%, but discussion of these data is beyond the scope of this paper. In the absence of precise knowledge of the absolute radiometric uncertainty in the Viking data used here, we will use the 13% absolute uncertainty given by Klaasen et al. [1977]