ATMOSPHERIC CARBON DIOXIDE AND RADIOCARBON: II 115 



of the best available radiocarbon data. At the same time, we were disappointed 

 to find that available radiocarbon data are not sufficiently accurate to establish 

 the rate of oceanic uptake of C0 2 or, indeed, even to establish whether oceanic 

 or biological uptake is the more important in reducing the airborne fraction of 

 industrial C0 2 . 



To determine the CO2 uptake by the oceans requires that not only the 

 volume of the surface layer and the circulation time of deep water be known but 

 also the rate of turnover of water lying at moderate depths below the surface 

 layer. We were unable to establish this rate and hence the extent to which this 

 intermediate water has taken up industrial C0 2  As an expedient we treated the 

 volume of the surface layer as an adjustable parameter, ranging from 

 approximately the correct value to a rough upper limit of three times that 

 volume. Even with this upper limit, the model predicts that the land biota has 

 taken up an appreciable fraction of the industrial C0 2 of recent years. This 

 conclusion is weakened considerably, however, when we take into account the 

 combined uncertainty of all the observational data, including the amount of 

 industrial C0 2 actually produced and the radiocarbon content of surface ocean 

 water. 



It is ironical that mankind, by nuclear weapons testing, altered drastically 

 the distribution of radiocarbon in the atmosphere, biota, and surface ocean 

 water before scientists had had time to , jrfect methods to measure natural 

 radiocarbon. If no artificial radiocarbon had been produced, the atmospheric 

 Suess effect would today be so large that it could be measured to high precision, 

 and additional ' 4 C measurements of land plants and surface water could provide 

 data to establish within narrow limits the other critical parameters of a 

 geochemical model. 



We are instead severely limited as to what new data we can obtain. Our 

 model analysis, although inconclusive, at least defines what additional data may 

 still be sought with the best return for the effort. Owing to the complexity of 

 the land biota, we cannot expect to obtain precise estimates of change in 

 biomass directly. The best remaining approach is to make the optimum use of 

 the historical record of atmospheric x C in tree rings. Our model suggests that a 

 detailed time series from several carefully selected trees could improve the 

 control on model parameters. One may argue that the relatively high counting 

 errors for individual samples are a serious obstacle to detecting the small 

 historical Suess effect. This is not a valid objection, however, because these 

 errors are essentially random, and we need know only the global mean Suess 

 effect averaged over several years. This time— space average can be based on so 

 extensive a series of individual analyses that random errors contribute little to 

 the uncertainty. A less apparent obstacle is to eliminate systematic errors arising 

 from choosing trees that do not correctly reflect atmospheric variations. 

 Matching of long-time series from each tree would be the best means to remove 

 unrepresentative trees and at the same time would establish whether or not a 



