The preliminary measurements of atmospheric CO2 performed at Amsterdam Island between April an the end of 1980 showed two kinds of short-term variations. The first one, attributed to local effects of the photosynthesis of the island vegetation, does not seem considerably to affect the monthly mean CO2 concentrations. The second one has been attributed to long range transport phenomena. That corresponds to large scale meteorological perturbations and generally is correlated with variations of the atmospheric radon concentrations. Therefore, the existence of regional CO2 sources and sinks in the southern hemisphere can be felt at a very large distance. Consequently, isolated measurements of geographical gradients, could correspond to a transitory situation and should not be utilized for the determination of these sources and sinks.