26 R O G E R R E V E L L E A N D H A N S E. S U E S S the top layers of the sea were affected by a rise in temperature. Z) Decrease in the carbon content of soils: HUTCHINSON (1954) considers this the most probable additional cause of the Callendar effect. The increase in arable lands of about 4 x iols cm2 since the middle of the 19th century has resulted in a corresponding decrease of forest land of about 10%. Assuming that 70 % of all the soil carbon is in forests robably a considerable over-estimate), and !Eat cultivation reduced this by so %, the total decrease in soil carbon would correspond to 9 x 1ol6 gms of CO, which is 4 % of that in the atmosphere. At least four-fifths of this amount should have been transferred to the ocean. 3) Change in the amount of organic matter in the oceans. About 7 % of the marine carbon reservoir consists of organic material. The amount of organic matter must depend on the balance between the rates of reduction of CO, by photosynthesis and of production of CO, by oxidation. As pointed out previously, a change in the CO, content of sea water by a certain factor without a corresponding change in alkalinity changes Pco, by about 10 times this factor. Therefore a I % change in the concentration of organic material in the sea will change the partial CO, pressure and hence the atmospheric CO,,by roughly I %. During the past 50 years, the mcrease in marine carbon from absorption of industrial CO, of about 0.2 % might have increased the rate of photosynthesis without a corresponding change in the rate of oxidation per unit mass of organic matter, and thus decreased Pco,. An increase in the temperature of surface water might have increased the rate of oxidation per unit mass of organic matter, and hence increased Pco,. W e suspect that fluctuations in the amount of organic marine carbon might be an important cause for changes in the atmospheric CO, concentration. ERIKSON and WELANDER (1956) have discussed a mathematical model of the carbon cycle between the atmosphere, the land biosphere and dead organic matter, and the ocean, in which it is assumed that the rate of input of carbon to the biosphere is directly proportional, not only to the total size of the biosphere but also to the amount of CO, in the atmosphere. Their estimate of the land biosphere is 7 times larger than that given in Table 3. They conclude that most of that part of the CO, added by fossil fuel consumption, which has not been absorbed by the ocean, h a s y b a b l y gone into the biosphere. Erikson an Welander’s basic assumption that the amount of atmospheric carbon dioxide limits the growth of the terrestrial biosphere seems highly unlkely, in view of the fact that the principal photosynthetic production on land is in forests, where a deficiency of plant nutrients might be expected. In any case as HUTCHINSON (1954) has shown, the amount of carbon in the biosphere and soil humus has probably decreased, rather than increased, during the past century, because of the clearing of forests. In contemplating the probably large increase in CO, production by fossil fuel combustion in coming decades we conclude that a total increase of 20 to 40 % in atmospheric CO, can be anticipated. This should certainly be adequate to allow a determination of the effects, if any, of changes in atmospheric carbon dioxide on weather and climate throughout the earth. Present data on the total amount of CO, in the atmosphere, on the rates and mechanisms of CO, exchange between the sea and the air and between the air and the soils, and on possible fluctuations in marine organic carbon, are insufficient to give an accurate base line for measurement of future changes in atmospheric CO,. An opportunity exists during the International Geophysical Year to obtain much of the necessary information. Acknowledgements A paper on the same subject by James R. Arnold and Ernest C. Anderson appears in this issue of this journal; we are grateful to Drs. Arnold and Anderson for the opportunity of the problem before publication. W e are appy to note that these authors have discussin[ simultaneously and independently derived essentially the same conclusions as presented in this paper. W e hope that the somewhat different approach will make both contributions equally valuable to the reader. W e also wish to thank Dr. Carl Eckart for valuable discussions and Dr. Harmon Craig for much constructive criticism. Dr. Craig’s own careful analysis of the subject appears in a separate paper in this issue. Tellus IX (1957). 1