Simple models were constructed to assess with somewhat limited CO2 data the effects of both frequency of sampling and spatial distribution of sampling locations on the variance of estimates of interest to the global carbon cycle. The CO2 data for use in these models were obtained from seven air flask sampling locations, globally distributed in latitude but restricted to the longitude sector 80¿W to 170¿W, during the period 1977--1979. The results of analysis with these models show (1) that locations north of 30¿N are quite important in possibly providing information on carbon cycle exchange processes and that (2) improved sampling techniques including greater sampling frequency would be desirable for sampling locations in the southern hemisphere.