Study of the Earth

THE STRATIGRAPHIC PANORAMA 191surprisingly enough they have worked quite well. From a multiplicity oforiginally proposed systems certain ones have emerged and have been ac-cepted by stratigraphers in general because they have proved useful ref-erence standards for geological time on a world-wide basis. Naturally, theyare imperfect in many respects, and we might divide the chronostrati-graphic section quite differently if we had it to do all over again today.However, perish the thought that we should confuse the great work ofthe past by drastically changing our system divisions and their nomen-Onclature now! the contrary, I believe that we can continue to live quitecomfortably with what we have. Only we must understand what it isthat we really have inherited in our present systems—perhaps no more norless than rather arbitrarily chosen reference units for expressing geologicage—and we must sharpen and define them better in terms of type sec-tions of rock strata so that they may better serve this simple and most use-ful function, regardless of any other significance they may have.Concept of world-wide \"natural breaks\" Historically, most of the systems in their locality of origin contrastedstrongly with adjacent ones through outstanding differences in lithology,structure, or fossil content, and their authors beheved their boundaries tobe indicative of \"natural\" division points in earth history. {See Hedberg,1948.) Perhaps even more than the originators themselves, their imme-diate successors assumed these systemic units to represent distinct world-wide steps in earth history. Later work has now shown that most of thesupposedly sharp breaks in the type areas were actually due to localchanges in environment, or to unconformities and hiatuses that left gapsin the local stratigraphic sequence. Sedimentary sections have now beenfound elsewhere in the world filling in many of those gaps and completinga more orderly and continuous sequence of fossils, on a world-wide basis,than was originally believed to exist. However, following the impetus ofthese early ideas, the highly attractive belief has persisted that the systemsare \"natural\" units marked off by relatively abrupt world-wide changes inearth history and in the evolutionary sequence of life forms. Althoughsuch good fortune might seem to some of us little less than miraculous inthe light of the way the systems were originally established, still a co-incidence of the present systems with these so-called \"natural\" breaks isaccepted by so many, and would be so convenient if true, that its validitybecomes of critical importance in its effect on our whole stratigraphicphilosophy. {See good discussion in Dunbar and Rodgers, 1957, p. 302- 307.) The question is: should we identify rocks throughout the world witha particular system on the basis of a certain concept of world-wide eventsor characteristics supposed to mark the \"natural\" limits of that system?Or should we be content to identify rocks throughout the world with a

192 HOLLIS D. HEDBERGparticular system on the more empirical basis of time correlation to thebest of our ability with an established type of that system, regardless of anypreconceived notion of what its world-wide characteristics or relations toearth history should be?Evidence in diastrophic record for world-wide \"natural breaks\" Let us briefly consider first the evidence for world-wide diastrophism asa \"natural\" basis for the separation of the present systems, especially sincediastrophism might be expected to be reflected both in hthology and inthe sequence of life forms. In a masterful address before this Society 12 years ago, Gilluly (1949)shook the theory of periodic world-wide orogeny to its mountain-buildingroots. Gilluly (p. 562) noted that this theory had previously been ques-tioned by \"Shepard (1923), Berry (1929), Von Bubnoff (1931), Arkell(1933), Woodring (1938), Spieker (1946)\" and others. In a stimulatingpaper with the intriguing title, Palaeozoic, Mesozoic, and Kainozoic; ageological disaster, R. H. Rastall (1944, p. 163), speaking in England, hadsaid,\"Another point that is in much need of emphasis, particularly in this country,is that periods of orogeny are far from being world-wide. As a matter of factWhenthey are distinctly local. comparing the tectonic history of northwesternEurope with any area in the southern hemisphere, it would probably be muchnearer the truth to say that the major revolutions alternate rather than synchro-nize. ... It seems probable that orogeny was always going on somewhere inthe world and still is.\" More recently, Spieker (1956) has again vigorously and indignantlybelabored the theory, Arkell (1956, p. 641) has said,\"So far as our knowledge goes at present, it does not point to any master planof universal, periodic, or synchronized erogenic and epeirogenic movements.The events were episodic, sporadic, not periodic. There was no 'pulse of theearth'.\"Gignoux (1955, p. 248) in rejecting Stille's periodicity of orogeny has said,\"Considered in the light of global unity, geological phenomena have obeyedneither the baton of an orchestra nor the rule of a geometer.\"Likewise Tyrrell (1955, p. 411), after discussing Sonder's geological cycle(geosynclinal phase, transgression phase, orogenic phase, continental phase)says,\"The use of Sonder's scheme in this connection does not imply its full ac-ceptance, or that it is applicable all over the earth. ... It does not take into

THE STRATIGRAPHIC PANORAMA 193account the fact that the earth's crust consists of dissimilar units, with differenthistories, and most probably with differing and non-synchronous geologicalcycles. Moreover, geological cycles do not always coincide with the standardgeological eras. The great Russian platform, for example, has remained almostundisturbed since Precambrian time, and must have experienced a quite differ-ent cycle of events from those of geosynclinal and orogenic regions such aswestern Europe and eastern North America.\" King (1955, p. 723) favors the thought\"that orogeny and epeirogeny were episodic rather than continuous; that epi-sodes affected fractions of continents rather than whole continents or all con-tinents; and that episodes may be expressed in one place by compression of thecrust, in another by tension, at one place by orogeny, at another by epeirogeny.\"Hawkes (1958), Holmes (1960), Kulp (1960), and others have recentlybrought out evidence opposing the idea of abrupt periodic world-wideorogenies of a type that might serve as boundary markers for systems.Gastil (1960), however, does see a concentration of radioactive age datesat approximately 200 million-year intervals which he suggests representsa cyclic distribution of broad periods of crustal adjustment.Evidence in record of organic evolution for world-wide \"natural breaks\" Let us now turn to the evidence for \"natural breaks\" on a world-widescale in the record of organic evolution. The extremely valuable sym-posium on the Distribution of evolutionary explosions in geologic time,organized by Lloyd Henbest and published in the Journal of Paleontology(1952, p. 298-394), represented an effort to examine the evidence for ab-rupt world-wide changes in the paleontologic record which might appearto be the reflection of periodic world-wide diastrophism. I quote from Hen-best's conclusions (Henbest, 1952, p. 317):\"The evidence from stratigraphic paleontology . . . indicates that the diastrophictheory and some of its corollaries that are applied to stratigraphy representgreatly oversimplified and exaggerated inference. ... As we progress in fillinggaps in stratigraphic records and reconstructing the history of the earth, thelines of evolution will emerge as paramount and the gigantic synchronous pulseswill become more diffused and obscure as integral properties of earth processes.\" J. S. WiUiams (1954, p. 1604-1605) suggests,A\"World-wide 'natural' boundaries do not exist between geologic systems.'natural' boundary may exist in the limited area of the type section and inother areas, but not in all and probably not in most regions. Despite whatseems to the writer to be rather general recognition of this situation, it seemsthat most geologists more or less unconsciously think in terms of some kind ofa lithologic, faunal, or diastrophic break at systemic boundaries in local areas.\"

194 HOLLIS D. HEDBERG I quote also from Hawkes (1958, p. 318) who, summing up the evi-dence, including that from the Henbest symposium and the more recentCrust of the earth symposium, says,\"It is recognized that there is a sensitive connexion between faunas and theirenvironment. Changes in the temperature, composition, depth, circulation,muddiness, and the like of water have an influence on the nature, extent, andrate of speciation of marine animals. If, as seems highly probable, such environ-mental changes have been in progress continuously on the earth throughoutgeological time there would be no reason to expect organic evolution to showspecial advances at three arbitrarily chosen periods such as the supposedly brief—episodes of the Caledonian, Hercynian, and the Alpine revolutions and it isnow conceded that this is not supported by the evidence. The long-entertainedidea that bursts of evolution coincided with these three episodes is on the wayout.\" On the other hand, Schindewolf (1954), Newell (1956, p. 97), andmany others see abrupt paleontological changes at the boundaries of theerathems which are \"real, approximately synchronous, and are recog-nizable at many places in different parts of the earth.\" So the subject isstill far from being closed. Perhaps the most striking of all supposed \"natural breaks\" is that stillenigmatic discontinuity that separates the highly fossiliferous rocks andcomplexly organized life of the Phanerozoic from the barren or verysparingly fossiliferous rocks and primitive life forms of the Precambrian.Even here, however, conclusions must be reserved until we better under-stand its nature. In many places sedimentary strata go down from thebase of Cambrian fossils concordantly and in apparently continuous se-quence into the Precambrian. Vicissitudes in the preservation of fossilsand in environments of deposition probably make the \"base of the Olenel-lus zone\" or the \"base of Cambrian fossils\" far from a world-wideisochronous surface, and the supposed sharpness of the break may begreatly exaggerated. I have mentioned the Ediacara fossil occurrence in the Precambrian ofAustralia and I feel certain that before long other Precambrian finds willbe made which tell us still more about this presently mysterious abruptappearance of complex hfe forms in our rock record. In the meantime,there is every reason to push on with the chronostratigraphic classificationof the Precambrian to the best of our ability on whatever objective baseswe can find. I would say amen to Harold James' conclusion (1960, p.113) that,\"although the immense duration of time and the lack of diagnostic fossils areformidable obstacles to overcome, the problems of stratigraphy and correlationin the Precambrian can and must be solved. Despite the difficulties, the Pre-cambrian is not a world apart; it contains the same kind of rocks and reveals

THE STRATIGRAPHIC PANORAMA 195the same kinds of geologic processes known from the record of younger eras;the same principles apply and the same rules must be used. And as with rocksof the younger eras, stratigraphy and correlation are the very essence of under-standing the geologic record.\"Summation of evidence regarding world-wide \"natural breaks\" Passing on from the evidence of diastrophism and organic evolution withregard to world-wide \"natural breaks,\" we have already looked briefly atthe age significance of lithologic character, mineralogy, chemical composi-tion, changes in sea level, igneous activity, unconformities, and other rockcharacters and features affecting rocks, and have found only a little thereinon which to base a reliable world-wide chronology and much less on whichto base \"natural\" world-wide chronostratigraphic divisions. Chmaticchanges and any other phenomena that may possibly result from extra-terrestial influences might oflFer particular promise of sound world-wide\"natural\" time-stratigraphic divisions; but here again evidence for recog-nition must he in the strata themselves, and such evidence seems still tooscanty and conflicting to furnish us yet with any very practical basis ofclassification. Far be it from me to attempt to pass judgment on the mighty mass ofobjective data which must be evaluated to determine if, and if so whenand how, great world-wide \"natural breaks\" or revolutions may have takenplace in earth history with sufficient impact and abruptness to have lefttheir mark so clearly and so sharply in earth strata as to constitute thebasis for \"natural\" world-wide chronostratigraphic divisions; and far be itfrom me to attempt to pass judgment on whether such breaks, if they doexist, coincide with our presently accepted systems. If this is what youexpected me to do, you will be disappointed. All I can do is to rely on thework of those who have studied the evidence much more thoroughly andmuch more understandingly than I. I do not find agreement among them,but I do find so much well-reasoned judgment in opposition that I must conclude, regardless of what may eventually be proved, that it has not yet been demonstrated that world-wide \"natural breaks\" in the character and continuity of our strata exist at the scale of the presently accepted geo- logic systems, nor has it been demonstrated that the evidence at the boundaries of the present systems is such as to allow them to be con- sidered as the \"natural\" world-wide division points of the chronostratig- raphic scale. Views of USSR Stratigraphic Commission on chronostratigraphic classification Some of you here may agree with my conclusions, some I know will not; but for those of you who do, it may be well to show that I have by

196 HOLLIS D. HEDBERGno means been setting up a straw man to attack in your presence. Asevidence thereof, I present the 1960 conclusions of the official USSRStratigraphic Commission (Interdepartmental Stratigraphic Committeeof the USSR, 1960) purporting to represent the thinking of a vast groupof distinguished geologists on the other side of the earth: This says insummary: 1. The basic aims of stratigraphy are the age correlation of rocks andthe construction of a single reference scale of the geochronologic divisionsand corresponding stratigraphic divisions for the entire Earth, based onso-called \"natural\" steps or stages in the history of the physical develop-ment of the Earth and the evolution of organic life. 2. The subdivisions of this single, so-called \"natural\" stratigraphic scaleshould be based on the totality of all lines of evidence, and thereforeseparate kinds of stratigraphic classification such as lithostratigraphic, bio-stratigraphic, and chronostratigraphic are unacceptable. 3. The dividing of the history of the earth into the so-called \"natural\"steps is made possible by the irreversibility of geologic phenomena and bytheir periodicity, most clearly manifested in the alternation of long-con-tinued stages of slow and gradual evolutionary development with shorterstages of rapid transformation in the face of the earth concomitant withgreat rearrangements in the internal structure of the earth's crust. Thisperiodicity is also manifested in alternations of large transgressions andregressions of the sea, in corresponding changes in the course of organicevolution, in changes in the process of sedimentation and denudation, inchanges in paleogeography, in changes in igneous and metamorphic ac-tivity, and in large tectonic movements of wide geographic range. 4. The presently recognized systems are \"natural\" divisions generallycharacterized in their lower parts by a sequence from continental to marinetransgressive deposits and in their upper part by regressive deposits. Theirboundaries are frequently characterized by angular unconformities, strati-graphic breaks, abrupt changes of facies, and evidence of igneous activity.The systems are paleontologically distinctive and are successively markedby the appearance and wide development of new hfe groups of major rank. I think it is at once evident that the Russian approach to chronostrati-graphic classification differs considerably in fundamental philosophy fromthat which I have favored in this discourse. The Russians would start withthe assumption of the existence of world-wide \"natural\" steps in thehistorical development of the rocks of the Earth's crust, more or lessaccordantly reflected in all lines of stratigraphic evidence, and would thenaim to fit the earth's strata as well as possible into these steps. I wouldfavor a more objective approach, which would start by classifying strataindependently with respect to each of several kinds of stratigraphic cri-teria without the preconceived conclusion that any or all of these would

THE STRATIGRAPHIC PANORAMA 197show accordance with each other or with any \"natural\" over-all chrono-stratigraphic grouping of strata. If they did, I might rejoice, but if theydidn't I would feel no urge to make them do so. I would try to makechronostratigraphic units as significant of earth history as possible, but Iwould define them purely on the basis of type sections—stratotypes—andthen would extend them throughout the world strictly on the basis ofempirical time correlation with the stratotype, utilizing the sum total ofavailable evidence of any kind for determining time equivalence. TheRussian course is deductive; that which I favor is inductive. In a sense the Russian view retains somewhat the influence of the oldcatastrophists. Opposed to the idea of the essential continuity and uni-formity of change for the world as a whole is their idea of periodic world-wide breaks reflected in world-wide changes in strata and their contents.Opposed to the idea that the boundaries of the generally accepted systemsare rather arbitrary points is their view that they mark \"natural\" world-wide division points. Their concept is an attractive one and one which haslong appealed to stratigraphers. It should not be passed over lightly, andthe only point I wish to make is that I (and perhaps you) have not yetbeen convinced that there is valid objective evidence for it, nor have I(and perhaps you) yet been convinced that there are valid theoreticalreasons why it should be so. Certainly it is something that will eventuallybe demonstrated either true or false as our knowledge of world stratigraphydevelops; meanwhile, it is fortunate that both the Russians and we recog-nize the same standard systems and the need to establish for these, naturalor artificial as they may be, designated type sections (stratotypes). SUMMARY AND CONCLUSIONS Probably few if any of our systemic boundaries are adequately definedin type areas of the present time, and stratigraphic progress is impeded bygaps and overlaps at these boundaries and by futile controversies over theplacement of strata. Our crying need is for the careful designation anduniversal acceptance of limits in continuous, type, or reference sectionswhich can serve as standards for these systems. Without such definitionwe will have only endless argument and continued chaos. To obtain themfor these world-wide units will require world-wide co-operation—the estab-lishment of stratotypes by international commissions of qualified stratig-raphers whose findings will have the support of an authoritative inter-national geological body. With a universally accepted base of referenceonce fixed, these systemic boundaries can then be extended through timecorrelation by any means available to us—fossils, radioactive age determina-tions, and what not—around the world as best we can, but always witha standard reference to which we can return in case of doubt or contro-

198 HOLLIS D. HEDBERGversy. Our systems thus delimited may be for many of us simply standardand universally understood units of chronostratigraphic measurement andrecord, with nothing more holy about them than there is holiness in themile, the foot, or the meter. For others, they may be sacred chapters inearth history. It really matters not too much which, as long as we agreeon the means of definition and the means of extension. Overlaps we shall have to eliminate by assignment to one or the otherof the overlapping systems. The gaps, as they are filled, can also be as-signed to one or the other of the adjacent systems by adding to theirstandard reference sections, or even by giving new names to the inter-mediate strata if the situation justifies. The resolving power of our meansof age determination and time correlation is not so fine but that there willprobably always be a greater or lesser no-man's land of strata of uncertainsystemic assignment with varying distance away from the designated typeor reference sections, but such is to be normally expected, and there is noneed to strain facts to make sharp divisions of that which it is beyond ourpower to divide. {See also Williams, 1954.) The succession of strata onthe earth provides a spectrum of age similar to the spectrum of light, andthere is no more crime in referring to strata of uncertain age position be-tween type Jurassic and type Cretaceous as Jurassic-Cretaceous than thereis in referring to a color midway between blue and green wave lengthsin the hght spectrum as blue-green. In fact this is much more accurateand scientific than to insist on a label of either blue or green or Jurassicor Cretaceous when there is no more proof of one than the other byreference to the type. The picture of stratigraphy typified by the Russian viewpoint and alsoaccepted by many others is a beautiful one, and perhaps its validity mayone day be demonstrated to those of us who prefer for the present to pro-ceed more cautiously. On the other hand, I cannot but think that themore purely objective concept, which I have outlined, of a stratigraphythat aims to delineate the earth's strata the world over, just as they arefound, with respect to as many of their many features as may be of inter-est or utility, and then proceeds to the drawing of only such conclusionson earth history as the established facts justify is also a beautiful picture.I cannot help but repeat a quotation from the anniversary address of SirCyril Hinshelwood (1958, p. x) to the Royal Society, to which ProfessorHawkes (1959) has called attention, because it seems so appropriate tothe history of this matter of stratigraphic classification. It reads as follows: \"What the true seeker after knowledge in his heart desires is some sim-ple design which he feels must underlie the facts. . . .The search for prin-ciples which are aesthetically satisfying seems often frustrated by thecomplexity of nature; and the conflict between imagination and austereregard for truth seems often to result in the passage of scientific theories

THE STRATIGRAPHIC PANORAMA 199through three stages. The first is that of gross over-simphfication, reflect-ing partly the need for practical working rules, and even more, a too-enthusiastic aspiration after elegance of form. In the second stage thesymmetry of the hypothetical systems is distorted and the neatness marredas the recalcitrant facts increasingly rebel against conformity. In the thirdstage, if and when this is attained, a new order emerges, more intricatelycontrived, less obvious, and with its parts more subtly interwoven, sinceit is of nature's and not of man's conception,\" So, in conclusion, for my part I should prefer to continue to think ofthe stratigraphic record as a panorama—a picture that has changed con-tinuously and profoundly as it was unrolled, but one in which the changesin its various individual features overlap and intergrade and so blend thatin the ensemble it is all one scene. Some parts, to be sure, are concealedfrom us today, but when we fully know the crust of our earth, both on thecontinents and under the oceans, the chances are that in one place oranother the gaps in the rock record will be filled. Just as there is no sedi-ment deposited unless there has been erosion somewhere, so there is nosequence of rocks in the earth's crust that is not somewhere equivalentto an unconformity or hiatus, and no unconformity or hiatus in the localrecord that may not somewhere be represented by deposits. Just as localclouds may obscure the continuity of our geographic panorama as we flyacross the continent, so locally we have breaks and gaps in the rock record,but, for the earth as a whole, there still exists the continuous stratig-raphic panorama. REFERENCES CITEDAhrens, L. H., 1955, Oldest rocks exposed, p. 155-168 in Poldervaart, Arie, Editor,Crust of the earth: Geol. Soc. American Spec. Paper 62, 762 p.W.Arkell, J., 1956, Jurassic geology of the world: Edinburgh, Oliver and Boyd, Ltd.,806 p.Armstrong, H. S., 1960, Marbles in the \"Archean\" of the southern Canadian Shield:W. —Internat. Geol. Cong., pt. 9, p. 7-20.Bell, C, 1959, Uniformitarianism or uniformity: Am. Assoc. Petroleum Geologists Bull., V. 43, p. 2862-2865.Briggs, M. H., 1959, Dating the origin of life on earth: Evolution, v. 13, p. 416-418.Chillingar, G. V., 1956, Relationship between Ca/Mg ratio and geologic age: Am.Assoc. Petroleum Geologists Bull., v. 40, p. 2256-2266.Cox, Allan, and Doell, Richard R., 1960, Review of paleomagnetism: Geol. Soc.America Bull., v. 71, p. 645-768.Daly, R. A., 1909, First calcareous fossils and the evolution of the limestones: Geol.Soc. American Bull., v. 20, p. 153-170.W.DeRoever, P., 1956, Some differences between post-Paleozoic and older regional metamorphism: Geol. Mijn. (N.W. ser.), v. 18, p. 123-127.Dorf, Erling, 1960, Climatic changes of the past and present: Am. Scientist, v. 48, p. 341-364.Dunbar, C. O., and Rodgers, John, 1957, Principles of stratigraphy: New York, John Wiley and Sons, 356 p.Fairbridge, R. W., 1954, Stratigraphic correlation by micro-facies : Am. Jour. Sci., v.252, p. 683-694.

200 HOLLIS D. HEDBERGFaul, H., 1960, Geologic time scale: Geol. Soc. America Bull., v. 71, p. 6B7-644.Gastil, Gordon, 1960, The distribution of mineral dates in time and space: Am. Jour. Sci., V. 258, p. 1-35.Gignoux, M., 1936, Geologic stratigraphique: 2nd ed., Paris, Masson et Cie, 709 p. 1955, Stratigraphic geology: New York, Harper and Bros., 682 p. (English translation of 4th French edition, 1950, by Gwendolyn Woodford).Gill, J. E., 1957, Summary and discussion, p. 183-191 in Gill, }. E., Editor, The Proterozoic in Canada: Royal Soc. Ganada, Spec. Pub. 2, 191 p.Gilluly, J., 1949, Distribution of mountain building in geologic time: Geol. Soc. America Bull., v. 60, p. 561-590.Glaessner, M. F., 1960, Precambrian fossils from South Australia: 21st Internat. Geol. Gong., pt. 22, Pr. Int. Paleont. Union, p. 59-64.Hawkes, L., 1958, Some aspects of the progress in geology in the last fifty years. I: Geol. Soc. London Quart. Jour., v. 113, p. 309-322. 1959, Some aspects of the progress in geology in the last fifty years. II: Geol. Soc. London Quart. Jour., v. 114, p. 395-410.Hedberg, H. D., 1948, Time-stratigraphic classification of sedimentary rocks: Geol. Soc. America Bull., v. 59, p. 447-462. 1959, Towards harmony in stratigraphic classification: Am. Jour. Sci., v. 257, p. 674-683.Henbest, L. G., 1952. Significance of evolutionary explosions for diastrophic division of —earth history introduction to the symposium: Jour. Paleontology, v. 26, p. 299-318.Hess, H. H., 1955, Serpentines, orogeny, and epeirogeny, p. 391-408 in Poldervaart, Arie, Editor: Grust of the earth: Geol. Soc. America Spec. Paper 62, 762 p.Hinshelwood, Sir Gyril, 1958, Address of the President, Sir Cyril Hinshelwood, at the Anniversary Meeting, 30 November 1957: Royal Soc. London Proc, Ser. A, v. 243,p. v-xvi.AHolmes, Arthur, 1960, revised geological time-scale: Edinburgh Geol. Soc. Trans., v. 17, pt. 3, p. 183-216.Interdepartmental Stratigraphic Committee of the USSR, 1960, Stratigraphic Classifica-tion and terminology: Moscow, State Publishing Office, 2nd revised edition, editorA. P. Rotay, 60 p. (In Russian and in English).James, Harold L., 1960, Problems of stratigraphy and correlation of Precambrian rocks with particular reference to the Lake Superior region: Am. Jour. Sci., v. 258-A (Bradley Volume), p. 104-114.King, P. B., 1955, Orogeny and epeirogeny through time, p. 723-740 in Poldervaart,Arie, Editor: Crust of the earth: Geol. Soc. America Spec. Paper 62, p. 723-740.Kulp, J. L., 1960, The geological time scale: 21st Internat. Geol. Cong., pt. 3, Pr., Sec. 3, p. 18-27.Lowman, S. W., 1949, Sedimentary facies in Gulf Coast: Am. Assoc. Petroleum Geologists Bull., v. 33, p. 1939-1997.Miholic, S., 1947, Ore deposits and geologic age: Econ. Geology, v. 42, p. 713-720.Moore, R. C., 1955, Invertebrates and geologic time scale, p. 547-574 in Poldervaart,A.: Crust of the earth: Geol. Soc. America Spec. Paper 62, 762 p.Nanz, R. H., Jr., 1953, Chemical composition of Pre-Cambrian slates with notes on the geochemical evolution of lutites: Jour. Geology, v. 61, p. 51-64.Newell, N. D., 1956, Catastrophism and the fossil record: Evolution, v. 10, p. 97-101.Pettijohn, F. J., 1941, Persistence of heavy minerals and geologic age: Jour. Geology, v. 49, p. 610-625. 1957, Sedimentary rocks: 2nd ed.. New York, Harper's, 718 p.Poldervaart, A., 1955, Chemistry of the earth's crust, p. 119-144 in Poldervaart, A.,Editor: Crust of the earth: Geol. Soc. America, Spec. Paper 62, 762 p.Rastall, R. H., 1944, Palaeozoic, Mesozoic, and Kainozoic: a geological disaster: Geol.Mag.,v. 81, p. 159-165.W.Rubey, W., 1951, Geologic history of sea water: Geol. Soc. America Bull., v. 62, p. 1111-1147.Schindewolf, O. H., 1954, Uber die moglichen Ursachen der grossen erdgeschichtlichenFaunenschnitte: Neues Jahrb. Geol., Pal., v. 10, p. 457-465.

THE STRATIGRAPHIC PANORAMA 201Spieker, E. M., 1956, Mountain-building chronology and nature of geologic time scale: Am. Assoc. Petroleum Geologists Bull., v. 40, p. 1769-1815.Stubblefield, C. J., 1954, The relationship of paleontology to stratigraphy: Adv. Sci., V. 11, no. 42, p. 149-159.Tyrrell, G. W., 1955, Distribution of igneous rocks in space and time: Geol. Soc. America Bull., v. 66, p. 405-426.Weller, }. M., 1960, Stratigraphic principles and practice: New York, Harper and Bros., 725 p.Weeks, L. G., 1958, Habitat of oil: Tulsa, Am. Assoc. Petroleum Geologists, 1384 p.Williams, J. S„ 1954, Problem of boundaries between geologic systems: Am. Assoc. Petroleum Geologists Bull., v. 38, p. 1602-1605.Wilson, J. Tuzo, 1952, Some considerations regarding geochronology with special refer- ence to Precambrian time: Am. Geophys. Union Trans., v. 33, p. 195-203.



^ -^^ ' In the face of geological opinion of today the writer has been forced by the sheerCLIMA 1 ES weight of such evidence into accepting some form of Continental Drift to explain the AND \"face of the Earth\"; indeed a world without some form of crustal drifting would appearQjD Tp'pjjs^Q —to him as unreal as one lacking in biologicalCONTINENTS evolution. alex l. du toit, Our Wander- ^\"^ Continents (1937) A Theory of Ice Ages • MAURICE EWING AND WILLIAM L. DONNTHIS ARTICLE IS A PRELIMINARY REPORT OF NEW IDEASrelated to the origin of glacial climates; it is based largely on observationsmade during the last 20 years. Glacial climates pose two problems: (i) thestriking alternations during the Pleistocene epoch of glacial and interglacialstages and (ii) the even more striking change from the warm nonglacialchmate, which prevailed generally from the Permian to the Pleistocene, tothe cold and glacial conditions of the Pleistocene and Recent. If it is difficult to answer the second question, it is even more difficultto solve both problems on the basis of a single theory. The present study(J) offers an explanation for the alternations in climate during thePleistocene and proposes an explanation for the change from nonglacialto glacial climates. PLEISTOCENE GLACIAL AND INTERGLACIAL STAGES First we wish to develop the following principal points of the glacial-interglacial theory. 1) The melting of an Arctic ice sheet (such as exists at present) wouldincrease the interchange of water between the Atlantic and Arctic oceans,cooling the North Atlantic and warming the Arctic and making it ice-free,thus providing an increased source of moisture for the polar atmosphere. • From Science (June 15, 1956), pp. 1061-66. 203

204 MAURICE EWING AND WILLIAM L. DONN 2) Two factors would then favor the growth of glaciers: (i) increasedprecipitation over arctic and subarctic lands and (ii) changes in atmos-pheric circulation, the latter also resulting from the warmer Arctic andcooler Atlantic oceans. 3) The lowering of sea level would greatly decrease the interchange ofwater between the Atlantic and Arctic oceans, which, together with thecooling effect of surrounding glaciers, would reduce Arctic surface tem-peratures until abrupt freezing occurred. The fairly sudden reversal ofconditions favorable to glacial development would terminate the growthof glaciers abruptly. 4) As continental glaciers waned, the sea level would rise, causing anincreased transport of surface waters northward until the Arctic ice sheetmelted once again, completing the cycle. 5 Temperature changes in the surface waters of the Arctic and Atlanticoceans are thus the causes of, rather than the consequences of, the waxingand waning of continental glaciers. Abrupt change in Atlantic deep-sea sediments. Using radiocarbon meas-urements, Ericson et al. (2) and Rubin and Suess (3) have establisheddates as far back as 30,000 years ago for cores of deep-sea sediments. Usingpaleotemperature measurements based on the oxygen isotope ratio, Emili-ani (4) determined the temperature of the water in which pelagic Fora-minifera lived; his findings covered a time interval represented by thelongest cores available. Suess (5) then combined these two sets of resultsby plotting the radiocarbon ages for three deep-sea cores from the equa-torial Atlantic and Caribbean against the paleotemperatures of Fora-minifera tests. The resulting graph may be interpreted as showing atemperature decline of 1°C per 11,000 years for the interval from 90,000to 11,000 years before the present. Temperatures then began an abrupt in-crease at a rate of about 1°C per 1000 years from 11,000 years ago to a few thousand years ago. For the last few thousand years, temperatures haveremained about as high as the maximum value that was reached during all Pleistocene interglacial stages (see Emihani, 4, Figs. 2 and 3). The curve given by Suess for Core A 172-6 shows the abrupt tempera- ture change beginning about 16,000 years ago as compared with the change shown at 11,000 years in the other cores. Since the time scale for this core depends on only three points, considerable interpolation is required, and since Emiliani shows that a strong solution of carbonates occurred in a number of places, linear interpretation is likely to be unreliable In addition, faunal studies by many investigators show that a well- marked change in most cores indicates the termination of the Wisconsin age. In 1935, Schott (6) described the faunal break in many equatorial Atlantic cores, which was later noted in more recent longer cores—for example, by Bramlette and Bradley in 1940 (7), Phleger, Parker and Pierson (8), Ovey (9) and Schott {10, 11). Most recently, radiocarbon

A THEORY OF ICE AGES 205measurements on a number of cores described by Ericson and Wollin(12) from the Atlantic Ocean and Caribbean Sea show that this charac-teristic break {13), marking a change from cold to warm pelagic fauna,occurred 11,000 years ago (2). Cores taken in the Gulf of Mexico showa similar faunal break (J4), although as yet no quantitative temperatureor time measurements have been made. However, on the basis of thefauna change, it is possible to correlate the break in the Gulf of Mexicowith that described in the Atlantic Ocean. Cores from the Gulf of Mexico. Many sediment cores 30 to 40 feet longhave been taken on the Mississippi Cone {IS), which spreads over mostof the deeper part of the Gulf of Mexico from a vertex just off the Missis-sippi Delta. These cores, which show a layer up to 2 feet thick of fora-miniferal lutite, indicating warm water, overlie with a sharp transition alayer of essentially unfossiliferous lutite and silt, the bottom of whichwas never reached. Pending radiocarbon measurements, it seems reason-able to extrapolate the date of 11,000 years ago for the bottom of theforaminiferal layer. In cores taken on the nearby Sigsbee Knolls, whichrise 20 fathoms above the cone and above the reach of turbidity currents,Foraminifera of the cold-water type are abundant throughout the siltylayers {14). It is concluded that the extreme scarcity of Foraminifera inthe silty layer of the Mississippi cone is due to dilution from a great influxof elastics in Wisconsin time. The sediments give no evidence of anyclimatic change except that 11,000 years ago.Further, the beginning of rapid postglacial rise in sea level is indicatedby the abrupt decrease in clastic deposition on the Mississippi cone about11,000 years ago, when the drowned rivers retained their clastic sedimentsinstead of discharging them into the Gulf {IS).The result of a very recent investigation of turbidity current depositsmade by Heezen and Ewing {16) shows that an abrupt rise in sea levelmust have taken place about 11,000 years ago; this event is also taken tomark the close of the Wisconsin glacial stage. Conclusions from Atlantic and Gulf data. On the basis of these data, weregard 11,000 years ago as the date of the most recent significant tempera-ture change in the Atlantic Ocean, and also as marking the end of theWisconsin glacial stage. Additional breaks in isotopic temperatures (4)and faunal types (7, 8, 11, 12) occur in many of the Atlantic cores whichWepenetrate most of the Pleistocene. regard these as marking the transi-tion from earlier glacial to interglacial stages. The data show that this fairly abrupt increase in temperature in thesurface layers of the Atlantic Ocean about 11,000 years ago was the mostsignificant temperature change of the past 60,000 to 80,000 years, althoughsimilar changes occurred earlier in the Pleistocene. Because the beginningof glacial retreat which closed the last Wisconsin substage (Mankato)occurred about 11,000 years ago—for example, Fhnt (J7)—it appears that

206 MAURICE EWING AND WILLIAM L. DONNthe oceans became warm abruptly at the time retreat commenced andremained warm during retreat of the ice. We propose to show that the temperature of the surface layer of theocean, rather than external conditions, regulated the climate of the land.Specifically, we suggest that the alternating warm and cold stages of Pleisto-cene climate are the effects of fairly abrupt alternations between warmand cold conditions of the upper layer of the Atlantic and Arctic oceans.We are thus led to consider the cause of the pronounced fluctuations ofthe temperature of the Atlantic and Arctic oceans. Arctic Ocean cores and the influence of the Arctic. Three short cores(9, 14, and 16 centimeters long) and one long core (81 centimeters long)were taken in the deeper part of the Arctic Ocean by A. P. Crary andstudied by D. B. Ericson. The upper 20 centimeters of the Slcentimetercore contain abundant Foraminifera (Globigerina) , which decrease in thenext underlying 10 centimeters. The long section below 30 centimetersconsists of lutite, with granules and pebbles distributed sparingly through-out, and shows no Globigerina except for a few near the bottom. Severalradiocarbon measurements on the Foraminifera of the 16-centimeter core(18) gave an age from 18,000 to 23,000 years. Other cores, collected else-where in the Arctic by Crary, show thin sediment zones above the fora-miniferal layer. The presence of this upper foraminiferal zone, with no oronly thin overlying sediment, indicates an abrupt end of conditionsfavorable to the growth of Foraminifera and suggests that the ArcticOcean was open during the Wisconsin glacial stage. Further, the coarsefragments in the longer lower section of the core are attributable toice rafting under conditions that provided numerous icebergs free tomove in open water. This condition would be met by an ice-free Arcticduring Wisconsin time. The absence of Globigerina in the lower sectionof the core, except for the very bottom, can be explained by dilution byclastic sediments; and the presence of Foraminifera in the upper portioncan be explained by a decrease in the amount of elastics. This is analogousto the similar, well-founded observations on cores from the Gulf of Mex-ico that were described in a preceding paragraph. Based on the possibility that the Arctic Ocean can undergo both ice-Whenfree and ice-covered stages, we propose the following assumption:the Arctic Ocean is ice-covered, surface temperatures in the Atlantic in-crease and continental glaciers decline; when the Arctic is open, surfacetemperatures in the Atlantic decrease, and continental glaciers develop.With this assumption in mind, let us examine the process by which anice-free Arctic can bring about a glacial stage. Consequences of an ice-free Arctic Ocean. Given an open Arctic Ocean,the resulting large increase in absorption of insolation is well known. Thepresent circulation of water within the Arctic Ocean would be greatlyincreased (19) since the wind stresses would then be applied directly to

A THEORY OF ICE AGES 207the water. It is postulated that a marked increase in the interchange ofwater between the Arctic and Atlantic oceans would occur which wouldtend toward equalization of temperatures, cooling the Atlantic and warm-ing the Arctic. With the added effect of greater absorption of insolation,the ice-free condition of the Arctic Ocean would be maintained againstthe cooling effect of increased evaporation. The present mean temperature of the surface of the Arctic Ocean in—January is 35°C and in July, about 0°C, The vapor pressure of water (orice) at 0°C is 4.6 millimeters of mercury; at 3°C, 5.7 millimeters; and forice at— 30°C, 0.3 milhmeter. The change to an ice-free condition wouldnecessarily require an increase of winter ocean surface temperatures by atleast 35 °C. By extrapolating from the afore-mentioned data, it can beseen that the vapor pressure would increase by a factor of about 50 in thewinter season, while the summer ocean surface temperature and vaporpressure would be only a little increased. The principal effect of an openArctic is thus the providing of greater moisture during the long polarwinter. Although it is difficult to make a detailed analysis of the consequentchanges in atmospheric and oceanic circulations and the results thereof,the following general conclusions seem reasonable. 1) The increased evaporation from an open Arctic Ocean, particularlyin winter, would increase the precipitation over adjacent cold land areas,where lack of precipitation, rather than high temperatures, at presentprevents the growth of glaciers (for example, Stokes, 20). 2) The present polar high, with clockwise circulation, would be replacedby a low-pressure area with counterclockwise circulation because of thecontrast between the warm water and the surrounding colder land. Theresulting reinforced counterclockwise circulation of the Arctic Oceanwould tend to increase further the interchange of Atlantic and Arcticwaters. Judging from the present North Atlantic circulation. 3) The semipermanent North Atlantic low would be displaced roughly10 to 20 degrees southward. Increased zonal flow of air around the northernportions of this low would transport cool moist air over eastern NorthAmerica. At the same time, the presence of relatively warm water on allsides of the continents would promote the development of cool continentalhighs. 4) Winter and summer conditions over the continents would becomemore similar as a permanent continental ice sheet developed, with aresulting migration of the present polar front southward. 5) The contrast between the cold northern land areas and the warmeropen Arctic Ocean would result in a second, although weaker, polar frontalzone surrounding the polar low. A6) second belt of storms would therefore be present, providing some

208 MAURICE EWING AND WILLIAM L. BONNnourishment for growing glaciers at the northern margin, in addition tothe stronger nourishment at the southern margin. 7) The Atlantic Ocean would be cooled by lowered air temperaturesinduced by continental glaciers and by the abrupt cooling effect of in-creased interchange with the Arctic Ocean already mentioned. Fig. 1. Chase's map of hypothetical pressure pattern during a glacial age with an ice-free Arctic Ocean. Air mo- tion is clockwise about high-pressure areas (H) and counterclockwise about low-pressure areas (L). Continental highs correspond to present- day cold winter situa- tions. Chase has given a general confirmation of these views in an analysis J.(see Fig. 1) that followed a presentation of the ideas of this paper at acolloquium at the Woods Hole Oceanographic Institution {21). Hisanalysis is based on a study of extreme conditions shown on historicalweather maps and represents hypothetical mean isobars during a glacialstage having an open Arctic Ocean. Termination of a glacial stage. The glacial stage would be brought toan abrupt close by the development of a new Arctic Ocean ice sheet.Along most of an Atlantic Ocean profile, through either Iceland or Spitz-bergen, the sill depth between the Arctic and Atlantic oceans is between200 and 300 fathoms, and much of this is less than 50 fathoms, a depthgenerally accepted as the probable value for maximum Wisconsin decreaseWhenin sea level (see Fig. 2). the lowering of sea level reached about50 fathoms, a serious reduction in the interchange of water between theseoceans would occur. The reduced inflow of warm Atlantic water, togetherwith the cooling effect of the continental glaciers, would eventually allowa new Arctic Ocean ice sheet to form. The interchange of water wouldthen be reduced to even less than its present-day value, owing to the re-

A THEORY OF ICE AGES 209duced sea level. Thus, with the present radiation balance and the slowreturn to the current atmospheric circulation pattern, there would be notendency to remove the Arctic Ocean ice sheet until sea level returned toAat least its present position. reversal of the phenomena described asconsequences of an ice-free Arctic Ocean would consequently occur. Al-though gradual wastage of the continental glaciers would follow, thewarming of the Atlantic surface water would be more rapid because ofan abruptly diminished interchange with the Arctic. NORTH ATLANTIC PROFILE THROUGH SPITSBERGEN NORTH ATLANTIC PROFILE THROUGH ICELANDFig. 2. Two profiles across the North Atlantic Ocean indicating the re-striction of Arctic-Atlantic interchange that would result from a loweringof sea level about 50 fathoms (hatched area) across either section.Although a narrow deep zone exists west of Spitzbergen, this is knownto be a region of return flow from the north. Depths are in fathoms. If we consider the last glacial stage, we find that the studies of deep-seacores indicate that 11,000 years ago is the date of the end of Wisconsintime. Furthermore, the end of Wisconsin glacial conditions in the oceancorresponds with, and according to our theory, caused the end of theMankato substage on the continents. The glacial retreat between theTazewell and Mankato glacial substage maxima (for example, Flint 17)is only a minor fluctuation. The interglacial stage. The transition from glacial to interglacial con-

210 MAURICE EWING AND WILLIAM L. DONNditions in the Atlantic Ocean would be simultaneous with the time ofmaximum continental glaciation (neglecting minor fluctuations). As thecontinental glaciers wasted, their coohng effects would diminish. Theconsequent rise in sea level would slowly enlarge the cross section of thechannel between the Atlantic and Arctic oceans, providing an increase inthe northward transport of warm water, as is occurring at present. Even-tually a condition would be reached where the Arctic ice sheet woulddisappear. Considerable evidence has been given by Berezkin (J9) and by Crary,Kulp, and Marshall {22) which indicates that the Arctic Ocean has beenwarming recently. If that trend continues, open water over the entireArctic Ocean might occur within a few centuries, with consequent glacia-tion in northern latitudes. The presence of five temperature maxima, marking past interglacialstages, all at about the same temperature as the present (4) is strongevidence that an internal, self-regulating mechanism controlled the climateduring the Pleistocene. Emihani's graph of temperature versus time (4,Fig. 3) gives compelling evidence of oscillations of the system betweentwo quasi-stable states, with the significant external conditions remainingconstant throughout. This implies that present temperature is at themaximum value expected for an interglacial stage and that a decrease intemperature marking the onset of the next glacial stage may be expectedwithin some few thousands of years. In describing Pacific cores, Emiliani notes (4, p. 561) far less conspicu-ous temperature variations than in cores from the Atlantic. He explainsthis on the basis of greater vertical circulation in the Pacific, which seemsto be a rather ad hoc solution. The observed uniformity in Pacific tem-peratures, however, is an expected consequence of the theory proposedhere, in which strong temperature changes should be limited to the Atlan-tic and Arctic oceans. These observations also imply that some mecha-nism, involving only the Atlantic and Arctic oceans, is the most reasonablesolution for the Pleistocene climatic variations.Interpretation of Wisconsin glaciers. Our theory provides a new explana-tion for the Scandinavian and Siberian ice sheets. According to Flint (23),this sheet had a maximum thickness of about 10,000 feet, tapering offstrongly to the north and east, and less strongly to the south. The com-5°Wbined lateral extent was from about at the British Isles to about110°E at the Taimyr Peninsula. In view of the long continental path andthe prominent mountain barriers to the south and west, it seems difficultto imagine nourishment from storms arriving from these directions. How-ever, nourishment from the north would be provided by an ice-free ArcticOcean, but would be almost impossible with the ice-covered ocean usuallyassumed for the glacial age. The steep north slope of this ice sheet furthersupports the theory that the source of precipitation was to the north.

A THEORY OF ICE AGES 211 The Laurentide ice sheet, which covered nearly 5 milhon square milesat the Wisconsin maximum (23), extended westward to meet the Cor-dilleran glaciers, and eastward to a hne seaward of the present Atlanticcoast, with its southern boundary along the Missouri and Ohio rivers.The maximum thickness has been estimated by Flint {23) to be about10,000 feet. Although the northern boundary is not well known, it hasbeen assumed to be thin, but recent observations by G. Hattersly-Smith{24) give evidence of very severe glaciation on Ward Island and northernEllesmere Land at about 83°N. Here again, as with the northern and west-ern margins, it is difficult to explain the sources of nourishment on thebasis of present-day circulation and an ice-covered Arctic Ocean. However,the modified circulation described here provides for sources of precipita-tion from the Atlantic and the Arctic oceans, in addition to that comingfrom the south.Finally, it is well known that the areas of the northern hemispherecovered by Pleistocene glaciers are centered roughly at the northeasterncoast of Greenland, near the strait through which Atlantic and Arcticwaters interchange (see, for example, Flint, 23, plate 3). For the mostpart, glaciation in other areas was minor and was controlled directly bymountains. The distribution of Pleistocene glaciers again indicates thestrong influence of both the Arctic and North Atlantic oceans on Pleisto-cene continental glaciers.Climatic optimum. Evidence for open water in the Arctic Ocean inpost-Wisconsin time has been accepted by many writers (for example,Brooks, 2S) and attributed to the \"climatic optimum\" or \"thermal maxi-mum\" which many climatologists believe prevailed during the long in-terval from about 7000 to 2500 years before the present. The evidence isfound in part on islands and remote shores where correlation with estab-Welished chronology is difficult. suggest that this evidence for an ice-freeArctic pertains to the open Arctic we have postulated for Wisconsin timerather than to the climatic optimum.Other evidence for a climatic optimum is found further south on thecontinents, where it is correlated reliably with post-Wisconsin chronology,Although the climatic optimum is correctly dated here, we believe that itis a minor climatic fluctuation because it left no conspicuous evidence inmarine sediments (4, 12). Also, according to Fisk {26), there is no evi-dence of higher sea level than the present in the Gulf of Mexico duringall of post-Pleistocene time. Early man in the Americas. The facts about early man in the Americassupport the idea of an ice-free Arctic during Wisconsin time and henceduring earlier glacial stages. According to recent prevailing opinion—forexample, Eiseley, (27)—early man reached Alaska from Siberia in greatnumbers during late Wisconsin time. The usefulness of the accepted landbridge between Siberia and Alaska would have been very limited if the

212 MAURICE EWING AND WILLIAM L. BONNArctic Ocean had been ice-covered and the chmate far colder than at pres-ent. The Denbigh Fhnt complex (northwestern Alaska) has been esti-mated from geologic correlations to correspond to warm periods eitherearlier than 12,000 years ago or about 8500 years ago {28, 29). Based ondirect observations of the Denbigh flint work, Giddings (29) concludes:\"The Bering Strait region was already a culture center at the time ofdeposit of the Denbigh flint layer.\" Giddings further notes \"that most ofthe early flint techniques were distributed primarily on a broad band cen-tering at the Arctic Circle; they seldom strayed south.\" We believe that these observations refer to the time of the relativelywarm and open Arctic Ocean prior to 11,000 years ago. The imphcationof a long established culture in the arctic region conflicts strongly withthe conventional concept of a Wisconsin ice sheet continuous from theNorth Pole to the Ohio River. If the Arctic Ocean were open in Wiscon-sin time, we should expect evidence of settlements along most of theshores of the Arctic, contemporaneous with those in Alaska. Giddings (29)has already pointed out a similarity between cultures for the Denbighcomplex and settlements in northern Siberia. About 11,000 years ago, the break-up of the ice permitted such rapidmigration of Arctic population southward that the southern tip of SouthAmerica was reached in a few thousand years {27, 30). The initial avenuefrom Alaska was the high plains east of the Rocky Mountains {27, 31),which would have no mountain barrier if it commenced in the low un-glaciated area north of the Brooks Range and fronting on the Arctic Ocean. Following glacial retreat at the close of the Wisconsin stage, the route along the high plain east of the Rockies would have opened, while that from northern Siberia would have closed as sea level rose and iceformed in the Arctic Ocean. Thus, as early man migrated southward, continued migration from northern Siberia was cut off. INITIATION OF PLEISTOCENE GLACIAL CLIMATE Although the theory we have presented attempts to provide an explana- tion for the alternations of climate during the Pleistocene epoch, it cannotAgive an explanation for the initiation of cold Pleistocene climate. solu- tion to this problem is offered now. Reconsideration of the hypothesis of pole-wandering. Following the recognition of the extent and distribution of Pleistocene glaciers, many scientists sought an explanation of glacial climates in terms of major shifts in the positions of the poles. Much of the early work was summarized in 1883 by Hann (32), who believed that great secular changes in climate could only be accomplished by changes in the earth's axis of rotation. Kelvin and other physicists demonstrated that significant pole shifts would be impossible in view of the accepted evidence for high rigidity of the

A THEORY OF ICE AGES 213earth, thereby directing most subsequent studies toward alternative ex-planations. However, although they used different bases for their hypoth-eses, Koppen and Wegener {33, 34) and Milankovitch (35) never aban-doned this idea. In recent years attention has again been directed toward this hypothesisas it became clear that the earth could not be considered as a completelyrigid body. Thus, Vening Meinesz (36) concluded that, \"The forces caus-ing tectonic orogeny which are probably exerted by sub-crustal currentsmust have been amply sufficient for a shift of the poles,\" and he assumeda pole shift of many degrees as the basis of his explanation of the majorfracture pattern of the earth's crust. Runcorn (37) indicated that theearth's surface could undergo large displacements relative to the interioras a result of convection currents. In considering the direction of mag-netic fields indicated by studies of paleomagnetism, Runcorn {38) be-lieves that the variation in these fields could be fully explained by polewandering.Using a different approach, Jardetzky (39) recently reevaluated the workof Milankovich, concluding \". . . there was possible a slow secular dis-placem'ent of the crust in space, which was progressive during all geologicperiods. The cause of the rotation of the crust is the existence of a momentof centrifugal forces acting on the crust and due to the asymmetry of thedistribution of masses in the outer shell.\"The possibility that adequate forces exist to produce relative movementbetween the earth's surface and the interior has led us to reopen the ques-tion of the effects of pole wandering on secular changes in climate. Itshould be noted that the poles wander, according to present conception,in a relative sense. The differential movement between an outer shell andthe interior results in different points on the surface assuming the posi-tions of the poles.Climatic consequences of pole migration. The poles are presently lo-cated in positions of extreme thermal isolation, in marked contrast to theconditions that would prevail if both were in the open ocean. If the NorthPole were located in the North Pacific (for example, 35°N and 180°W)and the South Pole at the antipodes of this, in the South Atlantic Ocean,the free interchange of water with the polar regions would preclude for-mation of polar ice caps. The free interchange would further tend toAequalize temperature extremes both geographically and seasonally. re-sulting weak and uniform latitudinal temperature gradient would occur,in contrast to the present zonality. This kind of climate must have pre-vailed between the Permian and Pleistocene glaciations (and probablyduring the long intervals between other glacial periods), according toinferences made from the geologic record by all authorities. Based on different investigations, Kreichgauer {40), Koppen and Wege-ner (33), Milankovich {3S), Koppen {34), and Creer et al {41), have all

214 MAURICE EWING AND WILLIAM L. BONNplaced the North Pole in the North Pacific Ocean for a long intervalbeginning with the Cambrian. Although Milankovitch gave no dating, theother investigators estimated that the pole arrived in the Arctic OceanOnduring the Tertiary. the basis of the worldwide distribution of coralsMaof various ages, T. H. Y. (42) concluded that sudden displacementsHeof the solid earth shell with respect to the interior occurred. also lo-cated pre-Cretaceous pole positions at distances of more than 90 degreesfrom their present positions in order to reconcile the fossil record withthe appropriate climate, and concluded that abrupt shifts of the earth'scrust during the Tertiary then carried the poles to their present locations.From studies of rock magnetization, Hospers (43) concluded that polemigration since Eocene time could have amounted to 10 degrees. It is pro-posed here that the migration of the poles from an open-ocean environ-ment to the thermally isolated arctic and antarctic regions resulted in thechange from the warm equable climate to the glacial climates of thePleistocene. Assuming that the North Pole migrated into the Arctic Ocean, thecoohng effects of high latitudes would have become concentrated in thisregion owing to the isolation of the Arctic from the other oceans. In thesame way, the migration of the South Pole from the freely circulatingsouthern oceans to the Antarctic continent would have concentrated cool-ing effects over the land. Both polar regions became sources of cold\"polar\" air that contrasted strongly with the warm air from equatorialregions. The Pleistocene and Recent climates, characterized by markedzonality, were thus established. Growth of glaciers requiring for the mostpart only ample precipitation on cold continental regions (for example,Haurwitz, 44) was greatly favored by this climate. The Pleistocene typeof climate may thus be expected to continue as long as the poles remainnear their present thermally isolated positions. The motion of the poles was probably somewhat intermittent. If weconsider convection to be the mechanism producing this motion, orogenieswould be good indicators of convectional activity. The beginning of rapidpolar motion would coincide with the major orogenies at the end of theTertiary, and possibly also at the end of the Cretaceous. The climaticoscillations within the Pleistocene were far too rapid to be related to move-ments of the pole in and out of the Arctic region. As a consequence of the theories proposed, the principal alternationsbetween glacial and nonglacial stages would occur in the arctic. Relativelyminor changes would be expected in the antarctic, resulting primarily fromthe slight warming and cooling of the Atlantic Ocean. Despite these minorchanges, such as the present decrease in antarctic ice, the theory requiresa secular increase from zero thickness at the beginning of the Pleistoceneglacial epoch. There is some evidence that this has occurred. In many parts

A THEORY OF ICE AGES 215of the world, phases of high sea level are recorded by elevated beaches.Although there is some disagreement about correlations, many authors—for example, Zeuner (43)—identify five or six such beaches, at elevationsup to about 100 meters. Zeuner showed that a graph of beach elevationagainst time is approximately linear. He recognized that \"it seems prob-able that this straight line represents a more or less continuous drop ofsea level in the course of the Pleistocene on which the oscillations due toglacial eustasy were superimposed.\" It is now suggested that this appar-ently secular decrease in sea level, with attendant preservation of thebeaches showing successive decreasing sea level maxima, can be accountedfor by the secular growth of an antarctic ice cap. Numerous esitmates thatthe total decrease in sea level due to present ice caps is about 60 metershave been made; an additional decrease of about 50 meters can be at-tributed to thermal contraction of the sea water, if the present mean oceanWetemperature is taken as about 10°C below the Tertiary mean. canthus provide for a secular decrease in sea level of about 100 meters, whichseems to account for the highest of the elevated beaches. CONCLUSIONS The theories of the origin of the Pleistocene glacial climate and of theglacial and interglacial stages proposed here are in complete harmony withthe doctrine of uniformitarianism. No external influences or catastrophicevents are required to initiate or maintain these conditions. It is postulatedthat some mechanical process has caused the poles to migrate to positionsvery favorable for the development of glacial climates. The major changeswithin the Pleistocene are considered here to have resulted primarily fromthe alternations of ice-covered and ice-free states of the surface of theArctic Ocean. For the most part, this article pertains to the Pleistocene glacial epochand the warm interval between the Permian and the Pleistocene. Althoughlittle is known about possible glacial and interglacial stages during thePermian and Proterozoic glacial intervals, the initiation of such intervalscould have been a consequence of the same mechanism as that proposedhere for the initiation of the Pleistocene. The \"warm\" periods prevailingduring the long intervals between the times of glaciations before as well asafter, the Permian could also be explained, according to the theory pro-posed, as a consequence of the location of the poles in regions of freelycirculating oceans. The consequences of the ideas presented are that the Pleistocene chmatewill continue while the poles maintain their present locations and thatthe Recent epoch can be considered as another interglacial stage.

216 MAURICE EWING AND WILLIAM L. DONN REFERENCES AND NOTES1. This research was sponsored in part by the Research Corporation and the Engineer- ing Foundation. This article is Lamont Geological Observatory contribution No. 187.2. D. Ericson et al., in preparation.3. M. Rubin and H.E.Suess, Science 221,481 (1955); 123, 442 (1956) .4. C. Emiliani J. Geo/. 63, 538 (1955).5. H.E.Suess, Science i23, 355 (1956).W.6. Schott, Wzss. Ergeb, deut. atlantischen Expedition Meteor 3, 3 (1935).7. N. M. Bramlette and W. H. Bradley, US. Geol. Survey Profess. Paper 196 (1940).8. F. Phleger, F. Parker, J. Pierson, \"North Atlantic Foraminifera,\" Rept. Swedish Deep-Sea Expedition 1947-48. 9. C. D. Ovey, Roy. Meteorol. Soc. Centennial Proc. (1950).W.10. Schott, Goteborgs Kgl. Vetenskaps.- Vitterhets-Samhdll Handl. Sjdtte FoldgenSer.B6 (1952).1 1 . , Heidelberger Beitr. Mineral. Petrog. 4 ( 19 54 )12. D. Ericson and G. Wollin, Deep-Sea Research 3, 2 (1956) .13. In many of the deep-sea basin areas continuity in deposition is destroyed by theerosional or depositional action of turbidity currents; hence the chronology of thepast cannot be determined with precision. However, when it became possible totake, or select for study, cores undisturbed by such action [D. B. Ericson et al.,Geol. Soc. Amer. Spec. Paper No. 62 (1955), pp. 205-220], the pattern of deposi-tion where a layer with warm-water fauna overlies a layer with cold-water fauna wasAunmistakable. second complicating factor is introduced by animals, whose filledburrows frequently disturb sediment contacts. Allowance must be made for thiseffect in estimating the rate of change of sediment type.14. M. Ewing and D. Ericson, \"Studies of cores from the Gulf of Mexico,\" Prog. Rept.Lamont Geol. Observatory (1955).15. , Topography and Sediments in the Gulf of Mexico (American Assoc. Petroleum Geol., in press)16. B. Heezen and M. Ewing, Bull. Am. Assoc. Petroleum Geol., in press.17. R. F. Flint, Am. J. Sci. 253, 5 ( 19 5 5 )W.18. Broecker and J. L. Kulp, unpublished.19. V. A. Berezkin, Morskoi sbornik 4, 105 (1937).20. W.L.Stokes, Science 122,815 (1955).We21. are grateful to J. Chase for this general confirmation, which was sent in apersonal communication on 28 June 1955.22. A. P. Crary,J. L. Kulp, E. W.Marshall, Science 122, 1171 (1955).23. R. F. Flint, Glacial Geology and the Pleistocene Epoch (Wiley, New York, 1947).24. G. Hattersly-Smith, Arcfic 8, 1 (1955).25. C. E. P. Brooks, Climate through the Ages (McGraw-Hill, New York, 1949).26. H. N. Fisk, personal communication.27. L. C. Eiseley, Anthrobol. Soc. Washington IS (1955).28. D. Hopkins and J. Giddings, Jr., Smithsonian Institution Misc. Collections 121, 11 (1953).29. T- Giddings, Jr., Sci. American 190, 6 (1954).W.30. R. Hurt, Jr.,Am.Anfic/uify J8, 3 (1955).31. E. Antevs, \"The Quaternary of North America,\" in Regionale Geologie der Erde(Akademische Verlagsgesellschaft, Leipzig, 1941 )32. ].Hann, Handbuch der Klimatologie {\SS3).33. W. Koppen and A. Wegener, Die Klimat der geologischen Vorzeit (Berlin, 1924).34. W. Koppen, Mefeoro/, Z. 106-110 (1940).35. M. Milankovitch, Handbuch der Geophysik (Gebr. Brontrager, Berlin 1938), vol. 9, pp. 593-698.36. F. A. Vening Meinesz, Trans. Am. Geophys. Union 28, 1 (1947)37. S. K. Runcorn, Advances in Physics 4, 4 (1955) .38. , Endeavour 14, SS (1955).

A THEORY OF ICE AGES II 21739. W. S. Jardetzky, Trans. Am. Geophys. Union 30, 6 (1949)40. D. Kreichgauer, Die Aquatorfrage in der Geologic (Steyl, Kaldenkirchen, 1902).41. K. M. Creer, E. Irving, S. K. Runcorn, J. Geomagn. Geoelect. Kyoto, in press.42. T. Y. H. Ma, Bull. Geol. Soc. China 20, 343 (1940); Research on the Past Climate and Continental Drift (Privately published, Taiwan, 1952).43. Hospers, /. Geo/. 63, 1 (1953). J.44. B. Haurwitz and J. Austin, Climatology (McGraw-Hill, New York, 1949)45. F. Zeuner, Dating the Past (Methuen, London, 1950) A Theory of Ice Ages II • MAURICE EWING AND WILLIAM L. DONNIN A RECENTLY PROPOSED THEORY OF ICE AGES (J) WEformulated the thesis that (i) the Pleistocene Ice Age was initiated whenthe North and South poles migrated into the thermally isolated locationsof the Arctic Ocean and Antarctica, respectively, and that (ii) fluctuationsof glacial with interglacial climate during the Pleistocene epoch were con-trolled primarily by alternation from an ice-free to an ice-covered state ofthe surface waters of the Arctic Ocean. According to this theory, the localterrestrial conditions of thermal isolation and adequate precipitation,rather than broad, world-wide changes of terrestrial or extraterrestrial ori-gin, should be emphasized as the causes of Pleistocene glaciation. FURTHER NOTES ON THE NORTHERN HEMISPHERE Despite the feeling of some authorities that the effects of an openArctic Ocean would be quantitively insufficient to cause the amount ofglaciation that existed, the validity of the theory seems to be illustratedby present conditions in the Arctic and Antarctic regions. Thus, the unexplained glacial conditions which have continued inGreenland since the Pleistocene contrast very sharply with the presentice-free condition of northern Canada at the same latitudes. The signifi-cants geographic difference between Greenland and northern Canada istheir location with respect to the North Atlantic Ocean. As a result of thelocation of Greenland, there is enough moisture in its atmosphere to causesufficiently heavy precipitation of snow for the maintenance of glacialconditions, whereas the very scanty precipitation at the same latitudes in • From Science (May 16, 1958), pp. 1159-62.

218 MAURICE EWING AND WILLIAM L. BONNCanada results in the present lack of glaciers there. Also, the precipitationin the southern part of Greenland is much heavier than that in the north-ern part. Hence, an open Arctic Ocean during the Pleistocene seems tobe the only geographic condition which could have produced glacial con-ditions in northern Canada equivalent to those in Greenland today. Fur-ther, the effects of the combination of thermal isolation and adequateprecipitation can be seen from a comparison of present conditions in theArctic and Antarctic areas. The thick Antarctic icecap contrasts sharplywith conditions in the Arctic Ocean area, with the exception of those inGreenland. This can be explained by (i) the more complete thermalisolation of Antarctica than of the Arctic (the condition in the Arctic isthe result of the small interchange of water between the Arctic and At-lantic oceans, without which interchange the Arctic Ocean would have apermanently thick frozen cover); and (ii) the availabihty of moisture fromthe surrounding open oceans for snow precipitation on Antarctica. Suchprecipitation is very slight over the nearly completely landlocked Arctic. Greenland is similar to Antarctica in being thermally isolated, in beingbounded largely by open water, and in having an icecap equivalent inthickness to that of the icecap in Antarctica. Greenland is also similar toAntarctica in that its icecap was probably maintained with little changeduring the Pleistocene interglacial stages. The following observation re-corded by Charlesworth (2, vol. 1, p. 94) certainly supports this: \"ThePleistocene ice sheets had a maximum thickness . . . which significantlyenough is roughly that of the modern ice sheets of Greenland and Ant-arctica.\" In view of these conditions, we may expect the Greenland andAntarctic icecaps to be preserved, with minor fluctuations, as long as thePoles are located in their present positions. Thus, the present contrast between Greenland and northern Canadaand that between the Arctic and Antarctic regions, which result from localconditions, are comparable to contrasts between glacial and interglacialstages and make it a plausible conclusion that the latter are also theresults of restricted terrestial changes rather than of global or extrater-restial causes. Further evidence that there was formerly a source of precipitation in theArctic region lies in the position of the glacial divide, as determined fromindicators of ice movement and glacial rebound. On the basis of indicators,J. Tuzo Wilson (3) shows that the divide ran approximately east-westthrough central Canada, except where its course was controlled by topog-raphy. Earlier notions based on the highest elevations covered by ice, assummarized by Flint (4), and deductions based on the theory that thesource of nourishment was to the south (5), have placed this ice dividemuch further to the south. Also, detailed studies of ice movement inAlberta (6) show that movement was from the north rather than fromthe northeast, as had previously been supposed.

A THEORY OF ICE AGES II 219 Although Lee et al. have applied the above data on motion indicatorsto late glacial conditions (7), the data of glacial rebound suggest, also,that the North American Wisconsin ice divide lay in the vicinity of Hud-son Bay, thus giving independent supporting evidence for the existenceof a source of precipitation to the north of the terminal moraine line inthe eastern half of North America. The evidence for uplift northward isbest given by the elevated beaches of the present Great Lakes and theancient Lake Agassiz (2, vol. 2, p. 132; 4, pp. 250-251; 8). An upliftof six to eight inches per 100 miles per century is given for these areas.Uplift determined from elevated areas around Hudson Bay reaches amaximum of 1000 feet (2, vol. 2, p. 1321) and is continuing at present(9) at an undetermined rate. If the data from the Great Lakes region isextrapolated through Hudson Bay, it seems clear that a continuous thick-ening of ice occurred from the present hinge line northward to the Hud-son Bay region. Blake's study (13) indicates that rebound in Labrador is less than thataround Hudson Bay and supports the location of the divide shown byWilson. Glacial rebound on northern Ellesmere Island and Ward Hunt Island(about 83°N) varies from 100 to 200 feet along the shore to at least 600to 700 feet further inland {II), thus approaching the magnitude of theuplift at Hudson Bay, far to the south. This suggests and supports stillfurther the argument that there must have been a source of moisture inthe Arctic region. THE SOUTHERN HEMISPHERE We wish to elaborate here on the statement in part I of this discussion ( I ) to the effect that Pleistocene glaciation in the Southern Hemisphereregions other than Antarctica and the sub-Antarctic islands was mainlylimited to high elevations and was consequent upon the general coolingproduced by the much greater change in the Northern Hemisphere. From the excellent summaries of Charlesworth (2, vol. 1, p. 44; vol. 1,p. 1322) and Flint (4), it is noted that in South America glaciers ex-tended along the Andes, with a few gaps, from Cape Horn to SierraNevada de Santa Marta in Colombia. The glaciers broadened considerablyon Tierra del Fuego and on the plains east of the mountains in Patagonia.Pleistocene glaciation in Africa was confined to the Atlas Mountains ofFrench Morocco and the high mountains of Equatorial East Africa, bothof which areas have perennial snow fields today. In Australia, barely 150square miles in the Australian Alps were glaciated, and upland areas ofSouth Island (New Zealand), and of Tasmania (both south of 40°S)were extensively glaciated. Thus, except for Tasmania and the smaller region in Austraha, it is noted by Charlesworth (2, vol. 1, p. 44; vol. 2, p.

220 MAURICE EWING AND WILLIAM L. BONN1322) that Pleistocene glaciers were merely extensions of the glaciers thatremain today in New Zealand, the Andes, and Africa. Further, with theexception of Auckland and the Macquarie Islands, the sub-Antarctic is-lands also have existing glaciers which were more extensive in Pleistocenetime. A moderate lowering of the snow line in the Southern Hemisphere,which will confirm the foregoing reconstruction of glacial conditions, isexpected to result from the global cooling produced through the effects ofglacial and pluvial conditions in the Northern Hemisphere upon theradiation and heat budgets of the earth. A planetary decrease in the amount of absorbed insolation would re-sult from a rise in albedo of the Northern Hemisphere. This would followfrom the greater reflectivity of the ice in glaciated regions and of theclouds in the pluviated regions (the latter to be described in detail below.) The areas in the Northern Hemisphere which were ice-covered duringthe Pleistocene glacial stages and are ice-free today cover 10.7 millionsquare miles and are distributed around a latitude of 60°N. If we assumea mean cloudiness of 60 percent for this region in both glacial and non-glacial stages, the albedo in the remaining 40 percent would be raisedfrom 10 to 70 during a glacial interval. If 300 calories per square cen-timeter per day {12) is taken as the mean insolation received at thesurface at a latitude of 60°N, then the resulting decrease in insolationXavailable for absorption in the glaciated areas is 2.0 10^^ calories perday.Further, as will be described in detail below, about 12 million squaremiles of arid regions were well watered (pluviated) during glacial stages.The mean cloudiness of these regions, which are distributed around lati-tudes of 30°N and 30°S, is about 20 percent at present. If an 'increase incloudiness to 60 percent (a figure based on present equivalent areas) dur-ing the glaciopluvial stages is assumed, the albedo of these desert regionswould increase from 15 (for sand) to 80 (for clouds). If 470 calories persquare centimeter per day (IS) is taken as the mean surface insolation,the increased albedo would result in a decrease of absorbable insolation ofX A4 10^^ calories per day for the pluviated zones. total reduction ofX6.0 10^^ calories per day would thus occur for the combination ofglaciated and pluviated regions. (The difference in albedo between sea iceand rough water in high altitudes is so small that no significant changein this estimate would occur if the Arctic Ocean were open during aglacial stage, as postulated by our theory.) This is a significant percentageXof the direct insolation of 85 10^^ calories per day for the entire earth.It is noteworthy that the terrestial changes described seem capable of re-ducing the radiation budget of the earth without reliance upon extrater-restrial changes, and thus of producing a sufficient degree of cooling tobring about glaciation in the Southern Hemisphere. It is also noteworthy

A THEORY OF ICE AGES II 221that the pluviated regions are at least as important as the glacial areas inpromoting global cooling. In the foregoing calculations, no account hasbeen taken of the small effect of absorption in the atmosphere. Although it is generally admitted that uplift of land areas result in cool-ing of such regions, it should also be noted that a minor contribution tothe general cooling of the lands would also result from the glacial low-ering of sea level, since a change of 300 feet in sea level produces anaverage change in temperature of 1°F. ANTARCTICA The field evidence available for an estimate of the effect of the conti-nental ice budget on Antarctica during Wisconsin time is scanty andinconclusive at present. Similarly, the conclusions about the Antarctic icebudget that can be derived from existing theories of glaciation are quiteambiguous. Thus, the Antarctic icecap appears to be in approximateequilibrium at present with regard to height {13) and lateral extent. Yetevidence in the form of exposed glaciated mountain areas exists to indicatethat there was a former higher equilibrium stand of the icecap. Most au-thorities place this higher stand in Wisconsin time. Theories of glaciationrequire the assumption that, for the most part, the ocean and air surround-ing Antarctica were cooler during glacial stages. Such conditions wouldproduce a decrease in snow precipitation over Antarctica which wouldmore than offset the decrease in wastage which results from lowering oftemperatures. It is difficult to conceive of there having been glacial growthon frigid Antarctica during times when the surrounding environment wascooler than it now is. It seems more reasonable to suppose that formerhigher levels of the icecap were a result of growth during interglacialstages or even during the more recent climatic optimum. Possibly thecontinuing study of Antarctica will provide information for the dating ofthis higher stand; this is at present an unsolved problem. PLUVIAL STAGES The effect of the Pleistocene conditions of moisture in presently aridareas is second in importance only to the contemporaneous glaciation inhigher latitudes. The major desert areas, which are today uninhabitedbarren wastes, although they occupy a very large part of the temperatezones, were formerly fertile, well-watered lands (J4). These areas, whichwere often covered by very large lakes, include the Sahara and Arabiandeserts, the desert of central Asia, and the Austrahan Kalahari, the NorthAmerican, the Atacama, and the Patagonian deserts. No theory of glacia-tion and no investigation of Pleistocene glacial stages would be complete

222 MAURICE EWING AND WILLIAM L. DONNwithout an explanation of the pluvial stages and their relation to glacia-tion. Although there is a considerable amount of evidence which suggestsstrongly that pluvial and glacial conditions occurred simultaneously, themost positive evidence for this comes from Lake Lahontan in westernNorth America (IS). The Lahontan data refer only to the end of the last glacial stage, butvery strong evidence for glacial-pluvial simultaneity comes from observa-tions around the Caspian and Black seas. According to P. F. Fedorov,every transgression of the Caspian Sea which occurred during glacial ad-vances of Pleistocene time coincides, without exception, with a regressionof the Black Sea (16); hence, it seems that pluviation was contem-poraneous with glacial lowering of sea level throughout the Pleistoceneperiod. The predominant cause of present-day deserts is their location in eitherthe belt of subtropical calms (the horse latitudes) or in the trade windzone marginal to this belt; in these zones the dry air moves equatorward,becoming warmer and thereby able to carry increased amounts of moisture.A secondary cause is the location of these deserts on the lee sides ofmountains and along coasts bathed by cool ocean waters. Some desertareas are the result of a combination of all these causes. The higher stands of many lakes and rivers during the glaciopluvialstages were the result of the snows and meltwater of adjacent glaciers.But the largest of these pluviated regions, including most of the present-day deserts, were so remote from glaciated areas that the cause of pluvialconditions must be other than simple proximity to glaciers. The fact thatthere has been widespread rainfall in the past over broad areas which arenot only arid at present but which lie in climatic zones where conditionsare basically unfavorable for the formation of rain in significant amountsindicates strongly that a fundamental modification of the atmospheric cir-culation must have occurred during the glaciopluvial stages. In part I of this discussion ( I ) , the theory was advanced that the pres-ent north-polar high-pressure area is a reversal from a polar low, whichresulted from the contrast in temperature between the relatively warm,open Arctic Ocean and the surrounding cold, glaciated continents. Fur-ther, it was stated that the Iceland low-pressure area, which at presentweakens in summer and intensifies in winter, probably migrated southwardduring glacial stages. By an extension of the reasoning involved, it is pos-sible to construct a model of modified circulation which could account forthe pluvial conditions that have been described for the present majordesert areas. The critical changes in circulation, which have been de-scribed, in principle, by a number of investigators in the past, are outlinedbelow. 1) During a glacial stage, the Iceland \"low\" of the North Atlantic

A THEORY OF ICE AGES II 223would migrate southward and would maintain present winter intensity allyear as a result of the perennial temperature contrast between the coldglaciated continents and the relatively warm ocean. Increased storm in-tensity and frequency would therefore persist throughout the year, thepaths of the storms being deflected far to the south of the present paths, 2) At present, the belts of subtropical calms (the horse latitudes) arelocated at approximately 30°N and 30°S and show greater intensity overthe oceans in summer than in winter. During a glacial stage, this zonewould also migrate southward in the strongly glaciated Northern Hemi-sphere and would probably weaken over the oceans because of the per-sistence of cold conditions over the continents. 3) The combination of an icecap extending into the middle latitudes, orpresent Temperate Zone, plus the southward migration of both the Ice-land low and the horse latitudes would result in the southward displace-ment of the entire zone of the prevailing westerlies wind belt and henceof the entire belt of migratory cyclonic storms which predominate in thisbelt. These storms would consequently travel well into the regions which,at present, are deserts because they lie in the dry horse latitudes andadjacent areas, 4) Owing to the changes described above, polar air masses originatingover the icecaps in the middle latitudes would tend to meet the extremelymoist equatorial air much more frequently than at present, thereby gen-erating very intense storms which would yield the very high precipitationcharacteristic only of hurricanes today. 5) Although in general it would be cooler than at present as a result ofwidespread global cooling during a glacial stage, the low-pressure doldrumbelt would become relatively stronger through contrast with the very coldbelt of the middle latitudes. Further, this belt, now located north of theequator in the vicinity of continents, would probably be displaced some-what to the south of the equator as a result of the pronounced coolingof the northern continents. This would tend to increase the amount ofmoisture over the present desert regions of the low southern latitudes ofSouth America and Africa, thus increasing precipitation over the desertsof South Africa and the west coast of South America. 6) As a result of the present monsoon pattern in southern Asia and theIndian Ocean, the doldrums are located over Austraha during the north-ern winter. With glacial conditions existing over the northern continents,the present winter-type pattern would tend to become semi-permanent,bringing considerably more moisture and precipitation to Australia. It isa well-known fact, established from the fossil record [see Benson {17)],that, during the Pleistocene, large fauna with tropical affinities inhabitedAustralia. This and the pluvial conditions of central Australia can beexplained by the theory of the change in circulation; the small high-alti-tude glacier of southern Austraha could have existed in much the same

224 GILBERT N. PLASSmanner as do equatorial glaciers on the mountain areas of Africa and South America at present (J8). REFERENCES AND NOTES 1. M. Ewing and W. L. Donn, Science 123, 1061 (1956). Owing to an oversight, the value given (page 1066) for the thermal contraction of the oceans from a 10°C drop in temperature, during late Tertiary is about eight times too large. 2. J. K. Charlesworth, The Quaternary Era (Arnold, London, 1957). 3. J. T. Wilson, glacial map, in R. F. Flint, Glacial and Pleistocene Geology (Wiley, New York, 1957). 4. R. F. Fhnt, Glacial and Pleistocene Geology (Wiley, New York, 1957) 5. W.F.Tanner, Science 122, 642 (1955). 6. C. P. Gravenor and R. B. Ellwood, Research Council Alberta {Can.) Prelim, Kept. 57-1 (1957). 7. H. A. Lee, B. Craig, J. G. Fyles, Geol. Soc. Am. Abstr. (1957), pp. 90, 91. 8. L. V. Pierson and C. S. Schuchert, A Textbook of Geology (Wiley, New York, ed. 3, 1929 ) , p. 302; B. Gutenberg, Bull. Geol. Soc. Am. 52, 743 ( 1941 ) 9. J. T. Wilson, personal communication.W.10. Blake, Jr., Science 121, 112 (1955).11. G. Hattersley-Smith, Arctic 8, 26 (1955); R. L. Christie, Geol. Survey Paper Can., 56-9 (1957).12. F. A. Berry, E. Bolloy, N. Beers, Eds., Handbook of Meteorology (McGraw-Hill, New York, 1945).13. R. Revelle, H. V. Sverup, W. Munk, Abstr. in Trans. Am. Geophys. Union 36, 31 (1935).14. For numerous references supporting and documenting this statement see 2, vol. 2, eh. 41.15. W. S. Broeker and P. C. Orr, \"The radiocarbon chronology of Lake Lahontan and Lake Bonneville,\" Bull. Geol. Soc. Am., in press.16. P. V. Fedorov, \"Quaternary stratigraphy and history of the Caspian Sea.\" Isvest. Akad. Nauk S.S.S.R., Ser. Geol. No. J ( 19 57 )17. W. N. Benson, Rept. Australian New Zealand Assoc. Advance Sci. ISth Meeting (J92J), pp. 45-128.18. This article is Lamont Geological Observatory Contribution No. 288. Carbon Dioxide and the Climate • GILBERT N. PLASSSCIENTISTS HAVE LONG BEEN FASCINATED WITH THEproblem of explaining variations in the climate. For at least nine-tenths ofthe time since the beginning of recorded geological history, the averagetemperature of the earth has been higher than it is today. Between these • From American Scientist (July, 1956), pp. 302-16.

CARBON DIOXIDE AND THE CLIMATE 225warm epochs there have been severe periods of glaciation which havelasted a few milhon years and which have occurred at intervals of roughly250,000,000 years. Of more immediate interest to us is the general warm-ing of the climate that has taken place in the last sixty years. Theories of chmatic change are exceedingly numerous. Is it possible thatany one of these theories can explain most of the known facts about cli-mate? The most widely held theories at the present time call upon varia-tions in the solar energy received by the earth, changes in the amount ofvolcanic dust in the atmosphere, and variations in the average elevationof the continents. Although it is entirely possible that changes in each ofthese factors may have had an influence on the earth's chmate at par-ticular times and places, none of these theories alone seems able to ex-plain a majority of the known facts about world-wide climatic variations. Although the carbon dioxide theory of climatic change was one of themost widely held fifty years ago, in recent years it has had relativelyfew adherents. However, recent research work suggests that the usualreasons for rejecting this theory are not valid. Thus it seems appropriateto reconsider the question of variations in the amount of carbon dioxidein the atmosphere and whether it can satisfactorily account for many ofthe world-wide climatic changes. Because of the relatively low temperatures at the earth's surface andin the atmosphere, virtually all of the outgoing radiation from the earthto space is in the infrared region of the spectrum. Thus it is important toknow which constituents of the atmosphere absorb in the infrared. Thethree most abundant gases in our atmosphere are oxygen, nitrogen, andargon. However, none of these three gases absorb appreciably in the rele-vant spectral region in the infrared. If these were the only gases in ouratmosphere, our climate would be considerably colder than it is today. Theheat radiated from the surface of the earth would not be stopped in itspassage out to space with the result that the earth's surface would coolrapidly. Fortunately for us, three other gases occur in our atmosphere in rela-tively minute quantities: carbon dioxide, water vapor, and ozone. Unlikethe more abundant gases, all three of these rarer gases absorb stronglyover at least a portion of the infrared spectrum. The concentration ofcarbon dioxide in the atmosphere is about 0.03 per cent by volume; it isfairly uniformly mixed as high as accurate measurements have been made.Water vapor and ozone also exist in very small concentrations in theatmosphere, but the exact amount that is present varies with time andplace. The infrared absorption properties of carbon dioxide, water vapor, andozone determine our climate to a large extent. Their action has oftenbeen compared to that of a greenhouse. There the rays of the sun bringthe heat energy in through the transparent glass. However, the outgoing

226 GILBERT N. PLASSheat energy from the plants and other objects in the greenhouse is in theinfrared where glass is largely opaque. The heat energy is fairly effectivelytrapped inside the greenhouse and the temperature is considerably warmerthan outside. In a similar manner the temperature at the surface of the earth is con-trolled by the transparency of the atmosphere in the visible and infraredportions of the spectrum. The incoming radiation from the sun in thevisible portion of the spectrum reaches the surface of the earth on a clearday with relatively little attenuation since the atmosphere is transparentto most frequencies in the visible. However, in order to have a warm cli-mate, this heat energy must be held near the surface of the earth and can-not be reradiated to space immediately. The atmosphere is opaque orpartially opaque to a large range of frequencies in the infrared becauseof the absorption properties of the three relatively rare gases describedabove. Thus radiation emitted by the earth's surface cannot escape freelyto space and the temperature at the surface is higher than it would beotherwise. The atmosphere has just the same properties as the glass inthe greenhouse. The carbon dioxide theory states that, as the amount ofcarbon dioxide increases, the atmosphere becomes opaque over a largerfrequency interval; the outgoing radiation is trapped more effectively nearthe earth's surface and the temperature rises. The latest calculations showthat if the carbon dioxide content of the atmosphere should double, thesurface temperature would rise 3.6°C. and if the amount should be cutin half, the surface temperature would fall 3.8°C. The carbon dioxide theory was first proposed in 1861 by Tyndall. Thefirst extensive calculations were necessarily done by very approximatemethods. There are thousands of spectral lines due to carbon dioxidewhich are responsible for the absorption and each of these lines occurs ina complicated pattern with variations in intensity and the width of thespectral lines. Further the pattern is not even the same at all heights inthe atmosphere, since the width and intensity of the spectral lines varieswith the temperature and pressure. Only recently has a reasonably ac-curate solution to the problem of the influence of carbon dioxide onsurface temperature been possible, because of accurate infrared measure-ments, theoretical developments, and the availability of a high-speed elec-tronic computer. The fact that water vapor absorbs to some extent in the same spectralinterval as carbon dioxide is the basis for the usual objection to the carbondioxide theory. According to this argument the water vapor absorption isso large that there would be virtually no change in the outgoing radiationif the carbon dioxide concentration should change. However, this conclu-sion was based on early, very approximate treatments of the very complexproblem of the calculation of the infrared flux in the atmosphere. Recentand more accurate calculations that take into account the detailed struc-

CARBON DIOXIDE AND THE CLIMATE 227ture of the spectra of these two gases show that they are relatively inde-pendent of one another in their influence on the infrared absorption.There are two main reasons for this result: (1) there is no correlationbetween the frequencies of the spectral lines for carbon dioxide and watervapor and so the lines do not often overlap because of nearly coincidentpositions for the spectral lines; (2) the fractional concentration of watervapor falls off ver}' rapidly with height whereas carbon dioxide is nearlyuniformly distributed. Because of this last fact, even if the water vaporabsorption were larger than that of carbon dioxide in a certain spectralinterval at the surface of the earth, at only a short distance above theground the carbon dioxide absorption would be considerably larger thanthat of the water vapor. Careful estimates show that the temperaturechanges given above for carbon dioxide would not be reduced by morethan 20 per cent because of water vapor absorption.One further objection has been raised to the carbon dioxide theory: theatmosphere is completely opaque at the center of the carbon dioxideband and therefore there is no change in the absorption as the carbondioxide amount varies. This is entirely true for a spectral interval aboutone micron wide on either side of the center of the carbon dioxide band.However, the argument neglects the hundreds of spectral lines from car-bon dioxide that are outside this interval of complete absorption. Thechange in absorption for a given variation in carbon dioxide amount isgreatest for a spectral interval that is only partially opaque; the tempera-ture variation at the surface of the earth is determined by the change inabsorption of such intervals.Thus there does not seem to be a fundamental objection to the carbondioxide theory of climatic change. Further the temperature changes givenby the theory for reasonable variations in the carbon dioxide amount aremore than enough to cause noticeable changes in the climate. It is notusually appreciated that very small changes in the average temperaturecan have an appreciable influence on the climate. For example, variousauthorities estimate that, if the average temperature should decrease from1.5 to 8°C., the glaciers would again form over an appreciable fraction ofthe earth's surface. Similarly a rise in the average temperature of perhapsonly 4°C. would bring a tropical climate to most of the earth's surface.Before discussing in detail the carbon dioxide theory of climatic changeit is first necessar}' to study the various factors that enter into the carbondioxide balance, including the exchange of carbon dioxide between theoceans and the atmosphere.The largest loss of carbon dioxide from the atmosphere is due to theXprocess of photosynthesis which uses about 60 10^ tons per year. In asteady state precisely the same amount of carbon dioxide is returned tothe atmosphere each year by all the processes of respiration and decay ofplants and animals, provided only that none is permanently lost in the

228 GILBERT N. PLASSform of new coal, oil, and other organic deposits. At the present time, atXleast, this loss is very small (0.01 10^ tons per year) and can be neg-lected for all practical purposes. If this steady state of absorption andemission of carbon dioxide by the organic world is disturbed, for ex-ample, by a sudden increase of carbon dioxide in the atmosphere, it isknown that the amount used in photosynthesis would then increase. How-ever, after a very few years the processes of decay and respiration wouldalso have increased. Since an average carbon atom that has been used inphotosynthesis returns to the atmosphere from the biosphere in about 10years and virtually all of the carbon atoms return in 250 years, it followsthat the factors influencing the carbon dioxide balance from the organicworld would again be in balance in a very few years.The two most important contributing factors from the inorganic worldare the release of carbon dioxide from the interior of the earth by hotsprings, volcanoes, and other sources and the formation of carbonates inthe weathering of igneous rocks. They happen to be nearly in balance to-Xday. The first one adds and the second subtracts about 0.1 10^ tons peryear to the atmosphere. Thus it appears that as far as natural factors areconcerned, the amount of carbon dioxide taken out of the atmosphereis very nearly equal to the amount returned to it. The specific numbersgiven in this section are only order of magnitude estimates. The valuesgiven here are averages of some of the more careful estimates.Recently, however, man has added an important new factor to thecarbon dioxide balance. As first pointed out by Callendar, the combustionXof fossil fuels is adding 6.0 10^ tons per year of carbon dioxide to theatmosphere at the present time and the rate is increasing every year. To-day this factor is larger than any contribution from the inorganic world.Thus today man by his own activities is increasing the carbon dioxide inthe atmosphere at the rate of 30 per cent a century. The possible influenceof this on the climate will be discussed later.The oceans contain a vast reservoir of carbon dioxide; some of it isin the form of dissolved gas, but it consists mostly of carbonates in variousdegrees of ionization. From the known dissociation constants for sea water,it is possible to calculate the atmospheric carbon dioxide pressure that isin equilibrium with a given amount of carbonate in the oceans. At theXpresent time the carbon dioxide pressure is about 3 10\"^ atm.; thereX Xare 2.3 10^^10^^ tons of carbon dioxide in the atmosphere and 130tons of carbon dioxide and carbonates in the oceans. Thus the oceans con-tain over fifty times as much carbon dioxide as the atmosphere. If condi-tions should change, the oceans can add to or subtract from the amount inthe atmosphere. Kulp has recently shown from radiocarbon determinations that the deepocean waters at the latitude of Newfoundland were at the surface 1700years ago. This suggests that it may take tens of thousands of years for the

CARBON DIOXIDE AND THE CLIMATE 229waters of the deep ocean to make one complete circuit from the surfaceto the bottom and back. Only the surface waters of the oceans can absorbcarbon dioxide directly from the atmosphere. Since there is very littlecirculation between the surface waters and the ocean depths, the time forthe atmosphere-ocean system to return to equilibrium following a disturb-ance of some sort is at least as long as the turnover time of the oceans.Thus, if the atmospheric carbon dioxide amount should suddenly increase,it may easily take a period of tens of thousands of years before the at-mosphere-ocean system is again in equilibrium.Let us next examine some of the variations in the atmospheric carbondioxide amount in past geological epochs and their correlation with theclimate as deduced from the geological record. It is interesting that alarge number of these climatic variations can be explained simply andnaturally by the carbon dioxide theory.During the last glacial epoch of perhaps a million years' duration, fourdistinct periods of glaciation separated by warmer interglacial periodshave long been recognized. Recently Wiseman has studied the sedimentsof the deep ocean floor and has found evidence for ten distinct tempera-ture minima within the last 620,000 years. It appears that a fundamentalproperty of a glacial epoch is to have a climate that is continually fluc-tuating. The glaciers advance and then recede and repeat the cycle sev-Noeral times before the end of the glacial epoch. other theory of climaticchange seems able to explain in a simple and straightforward manner thesecontinual oscillations in climate during a million-year epoch of glaciation.In order to understand these oscillations let us consider the figure wherethe equilibrium pressure of the carbon dioxide in the atmosphere isplotted against the total amount of carbon dioxide in the atmosphere-ocean system. These curves were calculated as described above with theadditional assumption that the average temperature varies as predicted bythe carbon dioxide theory. Curves are shown when the oceans have 0.90,0.95, and 1.00 times their present volume in order to allow for the factthat the ocean volume decreases during a period of glaciation.XThe present value for the carbon dioxide pressure (3 10\"^ atm.) andthe total amount of carbon dioxide in the atmosphere-ocean systemX(1.32 10^4 tons) is marked with the letter \"P\" in the figure. Let ussuppose that a million years ago the carbon dioxide balance was upset andthat the total amount of carbon dioxide in the atmosphere-ocean systemXwas reduced 7 per cent to 1.23 10^^ tons and that it remained Gxed atthis new lower value throughout the ensuing glacial period. Let us furtherassume that if the average temperature should fall 3.8°C. that great icesheets would again form and cover sizable portions of the continents. Withthe reduced carbon dioxide amount the atmosphere-ocean system wouldfinally come to equilibrium at the point \"G\" in the figure. The new at-Xmospheric carbon dioxide pressure would be 1.5 10\"* atm. This would

230 GILBERT N. PLASS TOTAL CARBON DIOXIDE AMOUNT IN ATMOSPHERE -OCEAN SYSTEM CtONS)The equilibrium pressure of carbon dioxide in the atmosphere as a func-tion of the total amount of carbon dioxide in the atmosphere-oceansystem. Curves are shown when the oceans have 0.90, 0.95 and 1.00 timestheir present volume. The present value for these quantities is indicatedby the letter \"P.\" The line between \"G\" and \"N\" represents a possibleoscillation of the climate during a glacial period.reduce the surface temperature by 3.8°C. according to this theory; thiswould be sufficient to start a period of glaciation.Let us assume in agreement with the estimates of glacial authoritiesthat the glaciers contain about 5 per cent of the water of the oceans whenthe ice sheets have reached their maximum development. Since only smallamounts of carbonates are held permanently in glacial ice, the loss of thiswater by the oceans means that the oceans now contain too much car-bonate for their reduced volume. They release carbon dioxide, thus in-creasing the amount in the atmosphere.The atmosphere-ocean system again reaches equilibrium at the point\"N\" in the figure some tens of thousands of years later. This point rep-resents the equilibrium conditions when the ocean volume is 95 per centXof its present value and the atmospheric carbon dioxide pressure is 2.510\"^ atm. However, when the carbon dioxide pressure reaches this value,the average surface temperature rises to virtually its present value. It isthen too warm to maintain the glaciers and they start to melt. This processprobably takes thousands of years, but finally the oceans return to theirNoworiginal volume. the oceans do not contain sufficient carbonates fortheir increased volume; the atmosphere-ocean system is no longer in equi-librium. The oceans absorb additional carbon dioxide from the atmos-

CARBON DIOXIDE AND THE CLIMATE 231phere until after tens of thousands of years the system is again nearequilibrium at the point \"G\" in the figure. The reduced atmospheric car-bon dioxide pressure now causes the surface temperature to fall 3.8°C.and another ice sheet starts to form. This cycle continues indefinitely aslong as the total carbon dioxide amount in the atmosphere-ocean systemXremains fixed at 1.23 10^^ tons. The period for one complete cycledepends on the rate of circulation of the oceans, but may be very roughlyestimated as 50,000 years or more. The climate must continually oscillate from a glacial to an interglacialperiod until the total carbon dioxide amount is again increased by a changeWhenin one of the factors in the carbon dioxide balance. the total carbondioxide amount is reduced slightly below its present value, there is noOnstable state for the climate; it must continually oscillate. the otherhand, if some event should greatly reduce the total carbon dioxide amount(perhaps by 30 per cent or more), a permanent period of glaciation with-out these oscillations would be possible. In order to explain the variousstages in this cycle more clearly, specific numbers have been assumed. How-ever, it may be verified easily that none of the conclusions that have beenreached depend in a critical way on the particular numbers that werechosen. It should also be pointed out that, if there is sufficient time in thevarious stages of the cycle for the oceans to come to equilibrium with cal-cium carbonate, the form of the curves in the figure is somewhat changed,but none of the conclusions reached above is essentially altered.In addition to lower temperatures, increased precipitation is also neces-sary for the formation of extensive glaciation. Most theories of climaticchange have found it very difficult to explain this increased precipitation.For example, in the variable sun theory, a decrease in the sun's radiationreduces the surface temperature. However, this also reduces the energyAavailable to drive the general circulation of the atmosphere. decreasedcirculation presumably means decreased cloud formation and precipita-tion. In order to account for the increased precipitation an ingenious, butunconvincing modification of the variable sun theory states that glacialperiods result from an increase in the sun's radiations. The slightly in-creased average temperatures are suDDOsed to be compensated by thegreater precipitation. The carbon dioxide theory provides a simple, straight-forward explana-tion for the increased precipitation during a glacial epoch. One of theparameters that determines the amount of precipitation from a given cloudis the radiant loss of heat energy from the upper surface of the cloud.If this radiation loss increases, the temperature at the upper surface ofthe cloud decreases. This increases the temperature difference between theupper and lower surface of the cloud, which in turn increases the convec-tion in the cloud. Because of these more vigorous convection currents, itis more likely that rain will fall from the cloud. Thus on the average there

232 GILBERT N. PLASSis more rainfall from a given cloud if the radiation loss from its uppersurface increases.According to the carbon dioxide theory there is a smaller than normalamount of carbon dioxide in the atmosphere when glaciers are beginningto form. Not only the surface of the earth, but also the upper surface of acloud is cooler, since they can lose heat energy more rapidly to space.Recent calculations show that the upper surface of a cloud at a heightkmof 4 is 2.2°C. cooler when the carbon dioxide pressure is half the pres-ent value. Further the upward flux of radiation that strikes the lowersurface of the cloud is larger when the carbon dioxide amount is reduced;thus the lower surface of the cloud is warmer than before. Thus, thelarger temperature difference between the upper and lower surfaces of thecloud causes increased convection in the cloud; the level of precipitationshould increase appreciably. Thus, according to the carbon dioxide theory,colder and wetter climates occur together.There is considerable geological evidence that extensive outbursts ofmountain building occurred several millions of years before each of thelast two major glacial epochs. Again the carbon dioxide theory seems tobe the only theory that suggests a reason for the time lag between thesetwo events. During a major period of mountain building, tremendousquantities of igneous rock are exposed to weathering. In mountainouscountry the zone for the active disintegration of rock extends much fartherbeneath the surface than it does in flat country. The weathering of ig-neous rock changes it into carbonates, thus removing carbon dioxide fromthe atmosphere.The explanation of the time lag in terms of the carbon dioxide theoryis that large quantities of carbon dioxide are removed from the atmosphereby the increased weathering after a period of major mountain building.After some millions of years, the carbon dioxide content of the atmosphereis reduced sufficiently to bring on a period of glaciation. From estimatesof the increased weathering that occurs after the uplift of a mountainrange, it is found that the time lag is of the order of a million years.However, during an epoch of mountain building greatly increasedamounts of carbon dioxide must be released from the interior of the earthinto the atmosphere through volcanic vents and hot springs. Additionalmillions of years are required to use up this additional carbon dioxide bythe process of weathering. Thus the actual time interval between the on-set of an epoch of mountain building and the ensuing glaciation can beconsiderably greater than a million years, if large additional quantities ofcarbon dioxide are released from the interior of the earth. Indeed, if theseamounts are very large, weathering would be unable to reduce the atmos-pheric carbon dioxide content to a sufficiently low level to cause a glacialperiod. In fact some periods of mountain building have not been followed

CARBON DIOXIDE AND THE CLIMATE 233by extensive glaciation. Such theories of glacial change as the variation inthe amount of volcanic dust in the atmosphere and the change in theaverage elevation of the lands have found it difficult to explain why theglaciers do not form immediately after the uplift of a major mountainrange.During the geological history of the earth the amount of carbon dioxidelost from the atmosphere in the formation of coal, oil, and other organicOndeposits has varied widely. This loss is relatively minor today. the otherhand it would be especially large during a period such as the Carboniferouswhen there where extensive marshes and shallow seas. At the end of theCarboniferous the atmospheric-carbon dioxide content may have been re-duced to a very low level because of the tremendous quantities that hadbeen used in the newly formed coal and oil deposits. It is perhaps sig-nificant that the glaciation at the end of the Carboniferous may havebeen the most severe in the earth's history.Radiocarbon dating indicates that recent changes in climate have beencontemporaneous in both hemispheres. In the last fifty years virtually allknown glaciers in both hemispheres have been retreating. According tothe carbon dioxide theory, such changes in climate should occur at thesame time in both hemispheres. The exchange of air between the twohemispheres is relatively rapid. Even if the atmospheric carbon dioxidecontent should increase suddenly in one hemisphere through a variationof some factor that enters into the carbon dioxide balance, the amount inthe two hemispheres should again be equal in a relatively short interval onthe geological time scale, perhaps no more than a few decades. It shouldbe mentioned that it is possible to have glaciation in one hemisphere andnot the other even though the atmospheric carbon dioxide amounts arethe same. If one hemisphere has extensive mountain ranges and the otheris relatively flat, glaciers could spread from the mountainous region of onehemisphere, whereas they would be unable to form on the more levelland of the other hemisphere at the same average temperature. The carbon dioxide theory has given plausible explanations for the be-ginning of a glacial period and of the climatic oscillations that occur dur-ing a glacial period. What increases the total carbon dioxide amount suf-ficiently to bring a glacial period to an end? One possibility is that therock weathering is slowly reduced because of the increasing flatness of theland. In addition extensive glaciation probably reduces the rate of weather-ing for the fraction of the land surface that is covered by the glaciers.Thus, as the loss of carbon dioxide from the atmosphere for weatheringdecreases as a glacial epoch nears its end, the amount of atmospheric car-bon dioxide slowly increases until finally the surface temperature is toohigh to allow further growth of the glaciers. An extensive period of moun-tain building has occurred at intervals of roughly 250,000,000 years duringthe earth's history and a glacial period has followed in each case during

234 GILBERT N. PLASSthe time interval when sufficient carbon dioxide was removed from theatmosphere. What is the reason for the recent temperature rise that is found through-out the world? Will this trend toward warmer climates continue for someWetime? The carbon dioxide theory may provide the answer. have dis-6x10^cussed the burning of fossil fuels which is adding more than tonsper year of carbon dioxide to the atmosphere. If all of this extra carbondioxide remains in the atmosphere, the average temperature is increasingat the rate of 1.1 °C, per century from this cause. Since 1900 a carefulstudy of world temperature records shows that the average temperaturehas been increasing at roughly this rate. Of course, the agreement betweenthese numbers could be merely a coincidence.As the concentration of carbon dioxide in the atmosphere increases,there are two factors in the carbon dioxide balance that can change. Firstthe oceans absorb more carbon dioxide in order to come to equilibriumwith the larger atmospheric concentration. However, only the surfacewaters can absorb this gas and because of the slow circulation of theoceans, it probably takes at least ten thousand years for this process tocome to equilibrium. Whenever the carbon dioxide amount is increasingan upper limit for the amount absorbed by the oceans can be found at anytime by assuming that the atmosphere-ocean system is always in equilib-rium. The actual amount absorbed by the oceans will be considerably lessthan the amount calculated in this manner for at least several centuriesafter a sudden increase in the atmospheric carbon dioxide amount. In thefirst few centuries the surface ocean waters can absorb only a relativelysmall fraction of the additional carbon dioxide. The second factor that can change is the amount used in photosyn-Athesis. higher level of photosynthetic activity can be supported by theincreased carbon dioxide amount. As previously discussed, this process tem-porarily withdraws some of the additional carbon dioxide from the at-mosphere into the organic world. However, in a relatively few years theincreased rates of respiration and decay bring this process back into equi-librium and only a relatively small amount of carbon dioxide is perma-nently lost from the atmosphere. Thus it appears that a major fraction ofthe additional carbon dioxide that is released into the atmosphere remainsthere for at least several centuries. Even if there may be some question as to whether or not the generalamelioration of the climate in the last fifty years has really been causedby increased industrial activity, there can be no doubt that this will be-come an increasingly serious problem as the level of industrial activity in-creases. In a few centuries the amount of carbon dioxide released into theatmosphere will have become so large that it will have a profound in-fluence on our climate. After making allowance for industrial growth, a conservative estimate

CARBON DIOXIDE AND THE CLIMATE 235shows that the known reserves of coal and oil will be used up in aboutX1000 years. If this occurs, nearly 4 10^^ tons of carbon dioxide willhave been added to the atmosphere; this is seventeen times the presentamount. The total amount in the atmosphere-ocean system will have in-X Xcreased from 1.32 10^^ tons. Even if the atmos- 10^* tons to 1.72phere-ocean system is assumed to be in equilibrium at the end of thethousand year period, the atmospheric carbon dioxide pressure will beX3 10\"^ atm., which is ten times the present value; the correspondingincrease in the temperature from this cause will be 13.4°C. If it is furtherassumed that there would be sufficient time for the calcium carbonateto dissolve and come to equilibrium in the oceans, the atmospheric pres-Xsure will be 1.1 10\"^ atm. and the temperature rise 7.0°C. This lastfigure is a lower limit for the temperature rise that will occur because ofman's industrial activities; the actual temperature rise must be larger sincethere will be insufficient time for these various equilibria to be established.Our energy requirements are increasing so rapidly that the use of nuclearfuels will probably not change materially the rate of use of the organicfuels. Unfortunately it is difficult to obtain any direct evidence for the carbondioxide content of the atmosphere during past geological epochs. In factit is not even certain from direct measurements whether or not the carbonAdioxide content has increased in the last fifty years. plot of such meas-urements can be fitted nicely with a linear curve that increases by 10 percent in that time interval. However, the probable error for most of themeasurements is so large that this result is not very firmly established.Because of its importance to the climate, regular measurements of theatmospheric carbon dioxide content should be started at several differentcountr)' locations and continued for a number of decades. Since the at-mospheric carbon dioxide content varies somewhat with the past historyof the air mass and the time of year, a number of measurements are nec-essary in order to obtain a reliable average. The present predicted rise of3 per cent a decade could be easily observed with the present techniques ofanalvsis. As to the carbon dioxide content of the atmosphere at earlierperiods, only general discussions of the various factors that effect the car-bon dioxide balance can be given at the present time. It is possible thoughthat we will be able to calculate the carbon dioxide amount of a pastepoch from measurements of the ocean temperature and the rate of car-bonate deposition during that epoch together with further studies of theatmosphere-ocean equilibrium. There is some interesting evidence which suggests that the carbon di-oxide content of the atmosphere was once much larger than at present.It is known that plants grow more luxuriantly and rapidly in an atmos-phere that has from five to ten times the normal carbon dioxide amount.In fact carbon dioxide is sometimes released in greenhouses in order to

236 GILBERT N. PLASSpromote growth. Since plants are perfectly adapted to make maximum useof the spectral range and intensity of the light that reaches them from thesun for photosynthesis, it seems strange that they are not better adaptedto the present carbon dioxide concentration in the atmosphere. The sim-plest explanation of this fact is that the plants evolved at a time when thecarbon dioxide concentration was considerably higher than it is today andthat it has been at a higher level during most of the ensuing time. Highertemperatures than today during most of the earth's history would have re-sulted from this higher carbon dioxide content. In fact the geologicalevidence shows that warmer climates than today have existed for at leastnine-tenths of the time since the Cambrian period.Further evidence as to the carbon dioxide amounts in the past is pro-vided by the pH of sea water. There is a definite pH value associated witha given atmospheric carbon dioxide amount when the atmosphere-oceansystem is in equilibrium. Further, many marine animals are very sensitiveto the pH value, the higher marine animals being more sensitive in gen-pHeral than the lower. For example, herring are killed if the changesby more than one-half unit; lower marine animals such as sea urchins,diatoms, and algae cannot tolerate pH changes of more than one unit. This suggests that the pH of the oceans has not varied by more thanthese amounts since the time when these animals evolved or at most thatthe pH has changed extremely slowly so that these animals could evolveto live in the changed environment. However, even with the stringent re-quirement that the pH of sea water should not change by more than one-half unit, the atmospheric carbon dioxide amount can still vary by a factorof fifty and maintain equilibrium between the atmosphere and the oceans.Thus very large changes in the atmospheric carbon dioxide amount canoccur without influencing either marine or land animals; still larger varia-tions would even be possible over time intervals sufficiently long to allowthe animals to adapt to their new environment. All calculations of radiocarbon dates have been made on the assumptionthat the amount of atmospheric carbon dioxide has remained constant.If the theory presented here of carbon dioxide variations in the atmos-phere is correct, then the reduced carbon dioxide amount at the time of thelast glaciation means that all radiocarbon dates for events before the re-cession of the glaciers are in question.Variations in the concentration or distribution of any gas that absorbs inthe infrared portion of the spectrum can influence the surface temperaturein the same manner as we have already discussed for carbon dioxide. Ozoneand water vapor are the only two other gases that absorb in this region andalso exist in the atmosphere in sufficient quantities to have an appreciableeffect. Few suggestions have been made that relate variations in the con-centration of these two gases to the climate, since these changes do notseem to be related directly to definite geological factors. However, recent

CARBON DIOXIDE AND THE CLIMATE 237calculations have shown that variations in the distribution of ozone canappreciably change the surface temperature. Normally the ozone con-centration has a maximum in the stratosphere with relatively smallamounts at lower altitudes. Vertical air currents occasionally bring someof the ozone down from the stratosphere, thus greatly increasing the con-centration at lower altitudes. This is sufficient to increase the surface tem-perature from radiation effects by several degrees Centigrade.The relative humidity as a function of altitude is continually changingand a similar effect on the surface temperature exists for water vapor.These relatively rapid variations in temperature are superimposed on thosefrom carbon dioxide alone. The latter variations are relatively constantover long time intervals compared to the former. However, water vaporcan also have an effect over long time intervals, since the amount thatcan be held in the atmosphere decreases very rapidly as the temperaturedrops. During a glacial period the atmosphere has a smaller capacity tohold water vapor; for this reason the infrared heat energy from the earth'ssurface can escape more easily to space. Thus the influence of water vaporon the infrared absorption tends to reduce the surface temperature stillmore once a glacial period has started. The increased cloud amount dur-ing such a period also acts to reduce the surface temperature by reflectingthe incoming solar radiation back to space. Therefore the temperaturedecrease during a glacial epoch is probably somewhat greater than is cal-culated from the carbon dioxide effect alone. A very large number of different theories of chmatic change have beenproposed. As more evidence about past climatic change is obtained, eachtheorv has to meet continually more rigorous tests in order to explainthe known facts. Each of the major theories of climatic change predictsAa different temperature trend during the remainder of this century. com-parison of these predictions with the actual record at the end of the cen-tury will provide an important test of these theories. The variable sun theory predicts that the temperature will decrease forsome decades. The maximum of the eighty-year period in the sun-spotcycle probably occurred in 1947. Thus the total energy received from thesun including the ultraviolet should decrease for some decades when theOnrecords are averaged over the shorter periods in the cycle. the otherhand a continued increase in the average temperature could be justifiedby the variable sun theory only if measurement showed a correspondingincrease in the solar constant.Changes in the average elevation of the continents clearly cannot beused to explain any variations in the climate over a period of a few cen-turies. However, the volcanic dust theory predicts appreciably lower tem-peratures for a few years following volcanic activity that throws largequantities of dust into the atmosphere. The last such explosion was whenKatmai on the Aleutian Islands erupted in 1912. More volcanic explosions

238 GILBERT N. PLASSof this kind must occcur before sufficient data can be obtained to cor-relate with the predictions of this theory. At the present time it is entirelypossible that volcanic dust creates small perturbations in the climate whilethe general trend is determined by some other factor. On the other hand the carbon dioxide theory is the only one that pre-dicts a continually rising average temperature for the remainder of thiscentury because of the accumulation of carbon dioxide in the atmosphereas a result of industrial activity. In fact the temperature rise from thiscause may be so large in several centuries that it will present a seriousproblem to future generations. The removal of vast quantities of carbondioxide from the atmosphere would be an extremely costly operation. Ifat the end of this century the average temperature has continued to riseand in addition measurement also shows that the atmospheric carbondioxide amount has also increased, then it will be firmly established thatcarbon dioxide is a determining factor in causing climatic change. REFERENCESArrhenius, S., Phil. Mag., 4 J, 237 (1896)Callendar, G. S, Quart. J. Roy. Met. Soc, 64, 223 (1938); Weather, 4, 310 (1949).Chamberlin, T. C, /. of Geology, S, 653 (1897); 6, 609 (1898); 7, 545, 667, 751 (1899).Harvey, H. W., Recent Advances in the Chemistry and Biology of Sea Water, Cam- bridge University Press ( 1945 )Hulbert, E. O., Phys. Rev., 38, 1876 ( 1931 )Kaplan,L. D.,/. C/iem.P/zys., 18, 186 (1950); /. Meteor., 9, 1 (1952).Kulp, J. L., Scientific Monthly, 75, 259 ( 1952)Plass, G. N., /. Opt. Soc. Amer., 42, 677 (1952); /. Meteor., 9, 429 (1952); 11, 163 (1954); Bull. Amer. Met. Soc, 34, 80 (1953); Quart J. Roy. Met. Soc, 82, 30 (1956); 82, (in press) (1956); Tellus, 8, (in press) (1956).Plass, G. N. and D. I. Fivel, Astrophys. I, 117, 225 (1953); Quart. J. Roy. Met. Soc, 81,48 (1955);/. Meteor., J2, 191 (1955).Rubey, W. W., Geo/. Soc. Amer., BuZZ, 62, 1111 (1951).Shapley et al.. Climatic Change: Evidence, Causes, and Effects, Cambridge, HarvardUniv. Press (1953).Sverdrup, H. U., Johnson, M. W., and Fleming, R. H., The Oceans, Their Physics,Chemistry and General Biology, Englewood Cliffs, N. Prentice-Hall (1942). J.,Tyndall, J., Phil. Mag., 22, 169, 273 (1861 )Urey, H. C, The Planets: Their Origin and Development, New Haven, Yale University Press (1952).Wiseman, J. D. H., Proc. Roy. Soc London, A222, 296 (1954)

The Record of Climatic Changes as Revealed by Vertebrate Paleoecology • EDWIN H. COLBERT THE NATURE OF THE EVIDENCETHE RECORD OF CHANGES IN THE EARTH'S CLIMATE DUR-ing past geologic ages can be derived from various lines of evidence, andof these the history of the vertebrates, as revealed by their fossil remains,is of particular importance. In the first place, the evidence afforded by thebackboned animals through geologic time is important because it covers avast expanse of earth history. True, the record of the vertebrates is notquite as extensive in time as that of the invertebrate animals, but thedifference is not great. These two lines of evidence, combined with thoseof paleobotany and of stratigraphy, cover the whole time span duringwhich there is a record of life on the earth. The evidence of fossil vertebrates is important also because it coversthe widest possible range of ecological environments. Vertebrates havelived at all levels in the oceans, and at almost all levels on the land. Theyhave lived in continental waters and in upland environments of exceedingdryness. They have lived in the air. Because of these wide geologicaladaptations the vertebrates can be used in conjunction with all otherforms of life, both plant and animal, in the interpretation of past climates. The vertebrates are useful too because many of them have been verysensitive to climatic conditions, and therefore to the changes in climates.In this respect they may, of course, be compared with the invertebratesand with plants. In all forms of life the ecological tolerances vary greatlyfrom group to group, often down to the species level, so that when theevidence is properly evaluated and weighted it is frequently of the utmostimportance because of the implications that can be drawn from it con-cerning climatic conditions in past times. In the discussion of the problem at hand attention should be called tothe use of the word implications. It must be realized that the evidence • From Climatic Change: Evidence, Causes, and Effects., ed. Harlow Shapley (Cam-bridge, Mass.: Harvard University Press, 1953), pp. 249-71. Reprinted by permission©of the publishers. The President and Fellows of Harvard College. 239

240 EDWIN H. COLBERTof the vertebrates as it bears upon past climates is for the most part in-Wedirect evidence. assume that ecological conditions (and by extension,climatic conditions as well) were thus and so because of the presence ofcertain animals in the sediments. This assumption rests upon a basicprecept that must be accepted at the outset, if our interpretation of pastclimates upon the evidence of vertebrate paleoecology is to have orderand vahdity. The precept is that within limits the past is to be inter-preted in terms of the present. Of course this maxim is commonly acceptedin the fields of geology and paleontology—so much so that it is generallytaken for granted—but it is being emphasized here because it is of un-usual significance in the present study.In line with this assumption we suppose that animals of the past weremore or less similar to their modern relatives in physiologic requirements,and hence in their ecological tolerances. Consequently we assume thatfishes, amphibians, birds, and mammals of the past were more or less liketheir modern relatives in their life processes, in their habits, and in theenvironments within which they were contained. From this we are thenable to draw conclusions as to the environments and the climates in whichcertain animals lived by a comparison of environments and climates inwhich their modern relatives live.Of course such a method of interpretation must be used with greatcaution and with as much insight as possible, because animals are neveridentical in their needs. But it is reasonable to suppose, for instance, thatthe temperature limitations under which reptiles of the past lived wouldbe roughly similar to temperature limitations for modern reptiles; and toextend the argument to greater detail, it is reasonable to think that theecological conditions favorable to crocodiles in the past would be more orless similar to ecological conditions under which modern crocodilians live.On the other hand, the possibilities of closely related animals living instrikingly different environments must always be kept in mind. One needonly cite the limitations of modern elephants to tropical and subtropicalclimates as compared with the extensions of some of their closely relatedcousins, the extinct mammoths, into arctic realms. Examples such as thesecall attention to the need for careful interpretation of fossil evidence.In the attempt to interpret past climatic conditions from the verte-brates, evidence other than that of the fossils is used whenever possible.In many instances such evidence is lacking or, if present, is of little value,but in other cases there is material that, when taken in conjunction withthe evidence of the fossils, throws considerable light on the ecologic con-ditions in which the animals lived.The physical characteristics of the sediments are often helpful. Thusthe nature of the rocks may indicate whether the animals that they containlived in fresh or salt water, in muddv rivers and estuaries, or in clearlagoons, in swamps or uplands, in deserts or regions of much vegetation.