Direct measurements of the optical depth above the two Viking landers are reported for a period of covering the summer, fall, and winter seasons in the northern hemisphere, a time period during which two global dust storms occurred. The optical depth had a value of about 1 just before the onset of each storm; it increased very rapidly, on a time scale of a few days, to peak values of about 3 and 6 with the arrival of the first and second storms, respectively; and its steadily decreased shortly thereafter (> few days to few weeks) for both storms, with the decay occurring more rapidly during the initial period of decay. We have also carried out further analyses of observations of the sky brightness made with the lander cameras during the summer season to obtain improved estimates of other dust particle parameters, including the cross section weighted mean particle radius, several shape factors, and the imaginary indices of refraction. These results have been used to define the radiative properties of the suspended dust particles at solar wavelengths. Particle properties inferred by Toon et al. (1977) from their study of the Mariner 9 Iris data were employed to define the radiative properties of the dust at thermal wavelengths. In an effort to understand the effect of dust content on atmospheric temperatures and winds, we have incorporated the derived radiative properties of the dust into a 1D radiative convective model. When only radiation and thermal convection are taken into account, these calculations fail to reproduce the temperature structure determined during the descent of the landers to the surface. However, satisfactory agreement is achieved when allowance is made for the effects of vertical motions induced by large scale atmospheric dynamics. The sign and magnitude of the required vertical velocities are crudely consistent with those found by Pollack et al. (1976a) in their general circulation calculations. The diurnal temperature variations predicted by the 1D calculations for the observed optical depths are also in crude agreement with values inferred from orbiter and lander measurements. The 1D model predicts that the diurnal temperature change and daily mean temperature, averaged over the entire atmospheric vertical column, steadily increase as the optical depth of the dust increases to a value of several, and then subsequently change little. Thus, for small and moderate values of optical depth, there is a positive feedback between atmospheric dust content and both tidal and seasonal winds, but little feedback occurs for very large values of optical depth. A negative feedback exist between dust content and thermally driven topographic winds. The occurrence of these feedback effects are supported by several Viking observations. It is suggested that they play a role in the generation of certain types of local dust storms, in the development of local dust storms into global ones, and in the decay of global storms. At present, dust is being preferentially deposited in the north polar region due to scavanging of dust by CO2 ice precipitation. Using the results of this paper and those of Pollack et al. (1977), we obtain an estimate of the sedimentation rate of dust and water ice in the polar regions. These results imply a time scale of about 105 years to form individual laminae in the polar laminated terrain, a result consistent with astronomical theories of their origin. Analogous calculations imply that equatorial and mid-latitude regions are being eroded at a rate of about 7 m per million years.