102 BROECKER [CHAP. 4 



(4) Two of the world-ocean models with a cyclic Atlantic (case A, B and C) 

 reproduce the main features of the 14 C distribution including the high 14 C 

 concentration in the Arctic and the intermediate value for the Antarctic. 

 Moreover, the rate of cycling demanded for the Atlantic is consistent with 

 oceanographic estimates of the net northward transfer of surface water in that 

 ocean. The 14 C value for the northern North Atlantic is the key in determining 

 whether the deep waters of the Atlantic rise at the north or south end of the 

 ocean (the basic difference between the cyclic and noncyclic cases). Although 

 the available data suggest a high Arctic 14 C value, and hence a dominantly 

 cyclic Atlantic, they may not be typical of the source areas for deep water. 

 Thus more measurements are needed to resolve this point. 



6. Industrial C0 2 Effect 



As shown in Table IV, the models with either the thicker mixed layer or the 

 more rapid Antarctic exchange-rate yield predicted estimates of the magnitude 

 of the Suess effect (decrease in the atmospheric 14 C/ 12 C ratio resulting from the 

 combustion of fossil fuels) more consistent with the observed value of 2.4 + 0.7% 

 obtained from tree-ring data — 1950 a.d. (Suess, 1955; Fergusson, 1958; 

 Munnich, 1957). The predicted values are obtained by considering the industrial 

 CO2 addition to be a series of discreet 14 C withdrawals from the atmosphere 

 at various times between the years 1850 and 1950. The magnitude of each 

 withdrawal is estimated from fuel consumption data (see Revelle and Suess, 

 1957). Unfortunately, no upper limit can be placed on the mixed layer depth 

 through consideration of the Suess effect because, as the depth increases, the 

 predicted Suess effect approaches a limiting minimum value of 1.7%, a value 

 within the range of experimental uncertainty of the tree-ring data. In all these 

 computations exchange with the organic matter stored in soils has been con- 

 sidered negligible. This assumption is consistent with the radiocarbon ages of 

 200 to 2000 years found for soil organics (Tamm and Ostlund, 1960; Broecker 

 and Olson, 1960). 



7. 226 Ra Distribution 



Koczy (1958) has shown that the 226 Ra concentration in surface water 

 averages about one-half that in deep water. Assuming that radium is not 

 removed from the surface waters by either organic or inorganic particulate 

 matter, it is not possible to explain the radium distribution with any of the 

 models discussed above. As shown in Table IV these models all suggest that 

 the 226 Ra concentration in the mixed layer should be no more than 10% lower 

 than that in the deep ocean. The radium data suggest that water presently in 

 the mixed layer has not been in the deep ocean for about one radium half-life 

 (1600 years). In order to explain both the 14 C and 226 Ra data simultaneously, 

 a model similar to that in Fig. 9 would have to be adopted. In this model a 

 fourth reservoir is added between the surface and deep ocean reservoirs. 



