BULLETIN OF THE GEOLOGICAL SOCIETY OF AMERICAVOL. 62. PP. 399-416, 1 FIG- 1 PL. APRIL 1951

MEASUREMENT OF PALEOTEMPERATURES AND TEMPERATURESOF THE UPPER CRETACEOUS OF ENGLAND, DENMARK,

AND THE SOUTHEASTERN UNITED STATES

BY H. C. UREY, H. A. LOWENSTAM, S. EPSTEIN, AND C. R. McKiNNEY

ABSTRACT

Since the abundance of the O18 isotope in calcium carbonate varies with the temperature at which it isdeposited from water, the variation in abundance can be used as a thermometer. This paper discusses thefollowing problems: (1) the magnitude of the effect expected, (2) the mass spectrometer of high sensitivity,(3) the preservation of the record during geological time, (4) the constancy of the isotopic composition ofthe ocean, (5) the impossibility of using skeletons of air breathing animals, and (6) the temperature of marineanimals relative to their surroundings.

A Jurassic belemnite is used to show that the record has been retained since Jurassic times, and belemnitesof the Upper Cretaceous of the United States, England, and Denmark are used to determine the tempera-ture of this time at these localities. The temperatures are about 15°-16°C and indicate nearly uniform tem-perature over this latitudinal belt. Because of the possible variation in O18 content of the oceans and thelimited number of samples, these temperatures are regarded as preliminary.

CONTENTS

PageIntroduction 399Acknowledgments 400Fractionation of isotopes in chemical processes 400Problem of measuring paleotemperatures 401Measurements of isotope composition, and

temperatures of the Upper Cretaceous 406Discussion of the North American tempera-

tures 414References cited 416Addendum 416

ILLUSTRATIONS

Figure Page1. Cretaceous sections for localities studied... 413

Plate Facing page1. Seasonal growth of a Jurassic belemnite... 406

INTRODUCTION

Geologists have drawn many conclusionsfrom the purely qualitative evidence of geo-logical studies in regard to the past climaticconditions on the earth. These deductions arebased upon a great variety of evidence, and theability of the geologists to deduce as much asthey have in regard to these conditions excitesthe wonder and admiration of all the uninitiatedwho examine their work even casually. Palmtrees do not grown in Spitzbergen, accordingto all our evidence, unless the weather is mild—about the same as that of Florida. Heavyshells are deposited in warm water, and thinshells in cold water. Tree rings are definiteevidence of seasons.

During the last 15 years a number of peoplehave studied chemical properties of isotopic

substances. These studies include the demon-stration of differences in the chemical proper-ties of the hydrogen isotopes, the first calcula-tions of such differences in the case of otherof the lighter isotopes, extensive applicationsof such differences to the separation of isotopesand considerable evidence of such differencesfrom surveys of the abundances of the lighterisotopes in nature. From the standpoint ofthe natural abundances the early demonstra-tion that deuterium was not constant over thesurface of the earth, the demonstration byDole (1935, p. 2731; 1936, p. 268) that oxygenalso varies, particularly in the air, and thevariation of carbon from inorganic and organicsources by Nier (Nier and Gulbranson, 1939,p. 697; Murphy and Nier, 1941, p. 771) con-tributed to geology. Recently Thode, Mac-Namara, and Collins (1949, p. 361) have dem-

399

400 UREY ET vll.—TEMPERATURES OF UPPER CRETACEOUS

onstrated substantial variations of the sulfurisotopic abundances, and Thode, MacNamara,Lossing, and Collins (1948, p. 3008) havedemonstrated such variations in boron. Thepresent paper presents a way in which thenatural variations in the abundances of thelighter isotopes makes possible, in principleat least, quantitative measurement of paleo-temperatures.

ACKNOWLEDGMENTS

We are indebted to The Geological Societyof America and to the American PetroleumInstitute for grants which have made thiswork possible, and we wish to extend our mostsincere thanks for this help. During recentmonths the work has also been supported underTask Order XXVIII, Contract N6ori-20 ofthe Office of Naval Research.

FRACTIONATION OF ISOTOPES IN CHEMICALPROCESSES

Differences in the chemical properties of theelements, such as differences in vapor pressuresand in equilibrium constants of exchange reac-tions, are at least partially responsible for thesevariations in the lighter isotopes in nature.Quite certainly the variation of the hydrogenand oxygen isotopes in the fresh waters of theearth as compared with the sea is due to thedifference in vapor pressures of the varieties ofwater containing hydrogen and deuterium orthe three oxygen isotopes, oxygen 16, 17, and18. The variation in the abundance of thecarbon isotopes discovered by Nier cannotcertainly be assigned to this cause. He foundthat carbon 13 exists in organic matter in adecreased abundance as compared with thecarbon in carbonates. Whether the photosyn-thetic process concentrates the carbon isotopesby some equilibrium or near-equilibrium proc-ess is not certain. Since the oxygen isotopesof the air are not in equilibrium with theoxygen isotopes of the sea or fresh water, thiseffect is probably also produced by the photo-synthetic process or by the oxidation of organicmatter by atmospheric oxygen,1 and if a vital

selection of the isotopes occurs in the case ofone of these isotopes it may well do so inothers.

Variation in isotopic abundances which aredue to differences in thermodynamic properties—i e., processes which are in equilibrium or

TABLE 1.—CALCULATED FRACTIONATION FACTORSFOR O14 AND O18*

re co, - -H,otfl

0 1.022025 1.0176

Ratio 1.0044

SO, - -HjOW

1.02041.01571.0047

POi * -HiO(0

1.01041.00871.0017

1 However, Dole has recently proposed other ex-planations for this effect (1949, p. 77).

* McCrea (Doctoral Dissertation, Chicago, 1949)by attempting to include the effects of the potentialfields in crystals instead of assuming no such fieldsas was done by Urey (1947) secures slightly differ-ent values for the COa" — H2O constants.

very close to equilibrium—are understood inprinciple and in many cases in detail. On theone hand, it is necessary to know the quantumstates of these molecules, and on the otherit is necessary to have the exact theories ofstatistical mechanics in order to relate thesequantum states to the physical chemical prop-erties. For Molecules in the neighborhood ofordinary temperatures and above, it is un-necessary to consider the rotational quantumstates, since these are very close to classicalat these temperatures in the case of all mole-cules except those involving the hydrogen iso-topes. In this case' the entire difference in thechemical properties resides in the differencesin the vibrational energy levels. A great manycalculations covering exchange reactions of thelighter elements were made by Urey and Greffi(1935, p. 321) and have been recalculated byUrey (1947, p. 562), together with the mathe-matical formulae and spectroscopic data bywhich the calculations are made. In generaldifferences of a few per cent in the ratio of iso-topes in two chemical compounds which arebrought to equilibrium with each other withrespect to the exchange of isotopes are possiblefor elements even as high in the periodic systemas chlorine. The heavier elements can be ex-pected to show but very slight effects of thiskind. The calculations indicate that there

FRACTIONATION OF ISOTOPES IN CHEMICAL PROCESSES 401

should be differences in the ratio of the oxy-gen isotopes in water and carbonate ion, waterand sulfate ion, and water and phosphateion, (Table 1).

In these calculations the results are reportedfor equilibrium between liquid water and thecarbonate ion rather than between the gaseouswater and the carbonate ion. A slight differenceof a few per cent is expected in the isotopiccomposition, and slight temperature coeffi-cients are predicted The calculations show thatthis difference for the carbonate-water equilib-rium shoud amount to about 0.44 per centin 25°C, or 0.176%0 per degree C. The calcula-tions are not exact because experimental datado not supply us with the vibrational energylevels for those molecules containing oxygen 18,and hence the vibrational frequencies of thesemolecules must be calculated from the knownvibrational frequencies of the more abundantoxygen 16 varieties. The calculation dependsupon a choice of constants. In fact, in the caseof the carbonate ion five constants are neededto describe the potential energy function to theprecision of quadratic terms in the co-ordinates,whereas only four independent frequencies areavailable for the calculation of these constantsfrom the observed vibrational frequencies ofthe COp* ion. A similar situation exists in thephosphate and sulfate ions.

If a relation exists between the oxygenabundances in these compounds, and tempera-ture, and the water in which the compoundswere deposited, it should be possible to setup a thermometer for present-day organismsliving in the sea, if the isotopic composition ofthe water is known not to differ from the meanof the present seas, or, in case it does, if boththe isotopic composition of the carbonate andwater are determined. Thus it would only benecessary to establish an empirical temperaturescale and to measure the isotopic abundancesof oxygen in a shell deposited by an animal inthe sea in order to determine the temperatureat which deposition took place. This applica-tion may be of interest in some cases, as forexample in determining the temperature historyof an animal, but the process which we de-scribe is difficult, and wherever possible ordinarythermometers are likely to be considerablymore convenient. On the other hand, these

effects may be preserved during geologicaltime, and hence this method affords a possiblemethod for measuring paleotemperatures.

PROBLEM OF MEASURING PALEOTEMPERATURES

Since the predicted temperature coefficientfor the ratio of oxygen isotopes in calciumcarbonate relative to water is small, massspectrometers of very great precision and sen-sitivity are needed for this problem. It wouldbe desirable to measure paleotemperatureswithin less than a degree Centigrade, whichcorresponds to a change in the atomic weightof oxygen in the calcium carbonate of 7 X 10~7

atomic weight units, or a change in the abun-dance of oxygen 18 by 1 part in 2.5 million ofoxygen. We have set 0.01 per cent or better,corresponding to about 0.5°C or 1°F, as ourobjective in these experiments. As measure-ments of the ratio of isotopes have been con-ducted in the past, reports in the literature of0.1 per cent or 1%0 are about as high a pre-cision as has been attempted. Hence the firstproblem in the application of this method topaleotemperatures is the construction and oper-ation of very sensitive mass spectrometers.

Since we have examples of animals andplants laying down isotopes out of equilibriumwith their surroundings, as for example thecarbon of organic material, we may well askwhether the deposition of calcium carbonateby an animal or a plant occurs so as to leavethe oxygen isotopes in equilibrium with thewater from which deposition takes place; thatis, we may ask whether there is a vital effect.Probably the equilibrium deposition shouldbe closely approximated in many cases, forthe calcium carbonate deposited is constantlybathed with water from the surroundings andhence may exchange its oxygen with the sur-rounding water during the deposition process.However, one could quite readily imagine mech-anisms which would involve the deposition oforganic oxygen and thus invalidate these as-sumptions. The whole question could be an-swered only by experiment, so far as we canjudge from the present state of our knowledgeof the physiology of deposition of such sub-stances. The question of deposition of phos-phate on the other hand would appear to be

402 UREY ET A £.—TEMPERATURES OF UPPER CRETACEOUS

even more doubtful than the deposition ofcarbonate, since phosphate ion does not ex-change its oxygen with water in which it isdissolved, and hence perhaps the phosphatedeposited by animals generally does not haveits oxygen in equilibrium with its surroundings.Since phosphate is so intimately involved inbiochemical reactions, its oxygen may bebrought to equilibrium with water by them.Only careful experimentation can determine,the feasibility of the phosphate thermometer.2

Animals rarely deposit sulfate, and it is notan important substance for investigations ofthe type we envisage immediately. Silica isoften deposited by plants and animals as ahydrous silica. Since the relative proportionsof water and silica might change the relativeabundance of the oxygen isotopes, this doesnot appear to be a very promising line ofattack, though again experiments should bemade to determine whether there is any possi-bility of using a silicate thermometer. Of thepossible thermometers—that is, the carbonate,phosphate, sulfate, and silica the carbonatethermometer seems the most likely to givesatisfactory results. A combination of the car-bonate and phosphate would eliminate thewater, and hence the isotopic variation of wateralways present in brackish water and freshwater would be eliminated. Dr. Martin Stein-berg is investigating the possibilities of thisthermometer.

Assuming that the shell is laid down inequilibrium with the water, the next importantquestion is whether this composition is pre-served during geological time. In the firstplace, if radical metamorphosis takes place,as for example the conversion of limestone todolomite, no satisfactory record of the isotopicabundance can be expected, for the disolvedcarbonate should exchange with the water inwhich it is dissolved and then be redepositedwith the composition partially dependent onits original composition and partially on thewater which was the means of redeposition.

2 Dr. Mildred Cohen (private communication)informs us that she has found that some enzymatichydrolyses of phosphoric esters split the oxygenphosphorus bond though others do not. Hence amechanism is available which may establish equi-librium between water and phosphate in animalbodies.

Also, if aragonite is converted to calcite throughsolution in water, the record would be de-stroyed, but, if recrystallization without thepossibility of exchange occurs, the record mightbe preserved.

The second important method for destruc-tion of the temperature record is diffusionthrough the solid state. Shells as usually de-posited appear to be fairly compact, but micro-scopic examination shows that they consistmostly of very small crystals, often with definitecanals through the shell, as for example in thecase of the brachiopods or the echinoderms.Thus in considering the problem of diffusionit is not the gross size of the shell that is im-portant, but rather the small microscopic par-ticles, since in the course of geological timewater would probably pass through the shellby hydrodynamic flow, or diffusion in the liquidwould bring the isotopic composition of waterwithin the shell to that of water in which it isimmersed. The diffusion problem for consider-ing this situation exactly would be very com-plicated. Assuming that the initial conditionsare that (1) at the surface of the crystals thecomposition is that required for equilibriumwith the surrounding water, and that it remainsso throughout the diffusion process, and (2)that initially the composition of the crystaldiffered from that which would be in equilib-rium with the water in which it has beenimmersed during the time since the shell wasformed, and that it changes from this initialcondition in accordance with the diffusion lawsat some constant temperature, a calculationcan be made which relates the size of the crys-tal, the time, and the coefficient of diffusionto the composition of the shell remaining. Thefraction of the excess mean concentration overthat in equilibrium with the surrounding water,assuming that the coefficient of diffusion is thesame in all directions, is

Qo & (2m

irwhere a = — — , D is the diffusion coefficient,

/2

t the time, and / the length of the edge of thecube.

If we require that the crystal shall retain its

PROBLEM OF MEASURING PALEOTEMPERATURES 403

original composition within some arbitrary per- and his values and the values of D calculated

centage, the value of- can be estimated

from this equation. (If one uses different valuesof D for different axes, the result is the sameas though different lengths were used for the

TABLE 2.—VALUES or Q/Q0 FOR VALUES or a*

Di,na --w0.10.050.010.0050.0010.00010.00001

QQc

0.4610.5900.7950.8540.9150.9640.972

* We are indebted to Mr. L. G. Montet for thenumerical results given in this table.

direction parallel to the axis and for the twodirections perpendicular to this.)

Table 2 gives a few values of a for a fewvalues of Q/Qo-

The coefficient of diffusion of one ion can becalculated from the electrical conductivity pro-viding the current is carried by that ion andnot by the other. Joffe (1923, p. 461) concludedthat only the calcium ion carries the current,but to what extent the carbonate ion carriescurrent was not determined except that itscontribution was not detected. If we assumethat the carbonate ion carries some fraction, S,of the current, the coefficient of diffusion ofthe carbonate ion can be calculated from theresistivity of calcite from the equation,

RTSD =

where » is the number of charges on the ion—i.e., 2—, F is the faraday, 9650 EMU, C is thenumber of gram molecules of carbonate ionsper cm3, p is the specific resistance in absoluteunits, and S is the unknown fraction of theconductivity due to carbonate and assumedto be 0.01 here. If p is in ohms, this formulamust be multiplied by 10-". Curie (1889, p.385)3 measured the specific resistance of calcite,

from them are given in Table 3.Using the first two values of DcoT in the

table, we can estimate the time required fordiffusion to destroy the record. Thus at 20°Ca crystal of 1 mm. dimensions will retain 96.4

TABLE 3.—MEASURED SPECIFIC RESISTANCE OFCALCITE AND CALCULATED COEFFICIENTS OF

DIFFUSION

|| axis

±axis

PC

2010016015

100150

P

5.5 X 1014

5 X 10"3 X 1010

9.5 X 1016

2 . 4 X 1012

1.3 X 10"

°co7

4.4 X 10~23

4.8 X 10-20

8.1 X 10-'9

2.5 X 10~24

l . O X 10-20

1.9 X 10~19

per cent of the original concentration relativeto the equilibrium concentration of a changedenvironment for 2.3 X 1016 seconds (7 X 10"years), while at 100°C the same crystal willretain this record only 2.1 X 10" seconds(6.4 X 104 years). Crystals of this size will notlose the temperature record due to diffusionin times sufficiently long to be of considerableinterest, and the coefficients of diffusion maybe even smaller than those given in Table 3.

A shell may grow a centimeter per year or3 X 10~3 cm. per day within an order of magni-tude. For simplicity we assume that the con-centration of O18 with distance due to dailyvariation of temperature varies according to acosine law. Then the diffusion equation gives

x' / '

where we take the mean concentration as zero.If we can detect a concentration differenceequal to 0.5 of the original difference betweenmaximum and minimum, i.e., at x = 0 andx = I, we have

In 2D-A

P '

3 See Landolt-Bornstein Tabellen II, 1062 (1923).

Substituting D = 4.4 X lO"83 and I = 3 X10-', the time becomes 1.4 X 10" or half abillion years. If we ask for 0.75 of the original

404 UREY ET AL—TEMPERATURES OF UPPER CRETACEOUS

variation the time becomes fifty milion years.Thus even a daily variation of temperaturemight be partially preserved.

There is the question of whether the isotopiccomposition of the ocean has remained con-stant during geological time. Such a changewould be produced, if igneous rocks of one iso-topic composition are weathered and sediment-ary rocks of another composition are produced.At high temperatures the isotopic compositionsof all chemical substances should be the same,and, if this temperature was that of moltenlava or higher, differences in isotopic composi-tion of water and silicates should not be largeand at least they should be less than at lowertemperatures. The calculations referred toabove show that O18 concentrates preferentiallyin the solid phases and not in the water, andit can be expected that this will be true withrespect to the silicate rocks. Thus O18 shouldenter the sedimentary rocks and its concentra-tion in the oceans should decrease with time.Assuming that the ratio of concentration inwater and silicates is the same as for sulfateand water at 2S°C, namely, 1.016, and that600 g. of rock have been eroded for each 1000g. of water (Goldschmidt, 1933, p. 112), it iseasily shown that the water should havechanged its ratio of oxygen isotopes by 3.8%0

due to this cause. Dr. P. Baertschi has shownthat there are differences between ocean wateron the one hand and the Disco Island Green-land basalt containing native nickel-iron andother igneous rocks, and, on the other hand,sedimentary rocks of just the kind and magni-tude predicted, though the total differencebetween ocean water and sedimentary rocksis not 16%0 as predicted. These results willbe published elsewhere.

It is of interest to inquire as to the time atwhich these changes took place. The carbonateequilibrium should have taken place early inthe earth's history, and largely since then noimportant change in the O18 content of carbon-ates occurred. The shales which make up muchof the sedimentary rocks should exchange theiroxygen with water less readily than the car-bonates, but no information is available. Baer-tschi finds that sandstones have an increasedO18 content relative to the Greenland basaltand thus may have exchanged their oxygen

to some extent, though the difference is smallenough to be explained by equilibrium differ-ences in concentrations between the silica andother phases of igneous rocks. Perhaps it is aconclusion based on little more than prejudice,but we believe that it is most likely that themajor part of this change occurred before theCambrian. Until more data are available we willmake no correction for this effect.

If the water of the oceans has been steadilyproduced during the earth's history from plu-tonic activity, as maintained by some (W. W.Rubey, private communication), the entireproblem of the isotopic composition of theoceans must be considered from this point ofview and must involve measurements on theO18 and D concentrations in waters which mightbe expected to be juvenile, as from geysers,hot springs, and volcanoes. Such measurementshave not been made, but are desirable from thestandpoint of the work presented here as wellas for the evidence they might give relativeto this suggestion in regard to the origin of theearth's water.

The oxygen of the air has a higher isotopiccomposition than that of the sea and freshwater. The difference in isotopic compositionamounts to about 2 per cent for air oxygen ascompared with the sea. All animals use atmos-pheric oxygen, and hence produce water thatcontains an increased concentration of oxygen18. In the case of marine animals having gillswhich exchange water between the sea and theirblood stream, it can be expected that in spiteof the oxidation processes taking place withintheir bodies the water content of their bodieswill be very close to that of the sea. One canreason that this must be true, for the wateroxygen must pass through the gills approxi-mately as readily as does the elementary oxygen,and since the amount of water present at thesurface of the gills is larger than the amount ofoxygen by a factor of about 105, the rate oftransfer of the water oxygen must be muchmore rapid than the transfer of elementaryoxygen, and hence the isotopic composition ofthe blood stream must remain close to that ofthe sea. On the other hand, either marine orland animals that breathe air are suspect forour purposes, since atmospheric oxygen is highin oxygen 18 (i.e., plus ̂ 20%0). Land animals

PROBLEM OF MEASURING PALEOTEMPERATURES 405

drink fresh water with a decreased concentra-tion of oxygen 18, and animals of the sea eatfood containing a very considerable amount ofwater of the composition of water of the sea.Without knowing the relative importance ofthese factors it is difficult to have any confi-dence in any measurements made on air-breath-ing marine animals, and even less on landanimals. It would be possible to investigatethe blood of marine reptiles or mammals todetermine how nearly the oxygen 18 contentof their blood agrees with that of the sea.This has not yet been done.

The question arises as to whether marineanimals maintain their body temperature abovethat of the surrounding sea. They do not.The source of heat is the combustion of foodby oxygen received through the membranes ofgills, usually, or of the general body surface.The membrane through which the oxygen dif-fuses is the same as that through which heatis lost to the surroundings. The transport ofoxygen determines the rate of heat production,and this must equal the rate of loss by thermalconduction. Assuming the maximum concen-tration of oxygen in sea water—i.e., 6 ml. perliter—and that the oxygen combines with sugarto produce 56,000 calories per gram atom ofoxygen, and knowing the coefficient of diffusionof oxygen in water and the coefficient of con-duction of heat in water, we can set up the fol-lowing equation:

AC AJ56,0001)-=*-,

where D is the coefficient of diffusion, K thecoefficient of thermal conduction, AC the con-centration difference of oxygen between theblood stream and the water and not greaterthan that corresponding to 6ml. per liter, or.0005 gram atoms per cm3, A/ the differencein temperature between the water and bloodstream, and Al the membrane thickness throughwhich both diffusion and heat conduction takeplace. The quantity A/ is unknown, but is thesame for both processes and can be cancelledfrom the equation and the equation solved forAt. Substituting D = 2 X 10-' cm2sec-' andK = 1.42 X 10-' cal. cm.sec-'00, we find thatA/ = 0.38°C. This is the maximum differencein temperature possible under normal aerobic

conditions. Animals can expend energy forlimited periods of time under anaerobic con-ditions and thus raise their temperature abovethat calculated here, but such periods wouldseem to be negligible.

The preparation of carbon dioxide gas foruse in the mass spectrometer has constitutedone of the most difficult problems in this re-search. It is so intimately related to the prob-lem of the empirical temperature scale thatboth these problems and related ones will bediscussed in a separate paper. Impurities in thecarbon dioxide must not produce peaks atmasses 44 and 46 such that the ratio of inten-sities of these masses is changed by more thanone part in 104, which means an impurity inmass 44 of less than this amount and in mass46 of less than one in 2.5 X 10'. We are for-tunate in using a relatively clean part of themass spectrum. The carbon dioxide is preparedfrom calcium carbonate by the action of ortho-phosphoric acid to which phosphorus pentoxideis added to bring the concentration to approxi-mately 100 per cent. To this enough chromicoxide is added to give a slight permanent redtint, and the excess oxide is reduced withhydrogen peroxide and the excess decomposedby heating. The calcium carbonate sample isfirst heated in a stream of carbon dioxide-freehelium for 15 minutes at approximately 475°Cin order to destroy organic matter. Approxi-mately 9 cc. of the phosphoric acid and 50 mg.of the calcium carbonate are put in separatearms of a reaction vessel, evacuated, andbrought to 25°C in a thermostat. The acid isthen poured onto the carbonate. A vigourousevolution of carbon dioxide results, sometimeswith a little froth which breaks during half anhour in the thermostat. The gas is then trans-ferred to sample tubes and analyzed on themass spectrometer. All stop cocks and taperedjoints are greased with Apiezon N cock grease.We have never detected any effect of impuritiesdue to this material.

With this procedure reproducible analysesare secured for both recent specimens and fos-sils. Using the shells of present-day marineanimals grown at known temperature and cor-recting for the variation of isotopic compositionof the various sea waters, the following equa-

406 UREY ET AL—TEMPERATURES OF UPPER CRETACEOUS

tion for temperature is secured (Epstein,Buchsbaum, Lowenstam, and Urey, 1950):

t = 11.88 - 5.918,

where t is measured in °C, and & is the differ-ence in %0 between the ratio of O18 to O16 ina standard working gas, and this ratio for thesample, i.e.,

I Rumplt _ \ltandard /

1000,

where R,amru and R,t,ndart are the ratios ofC016OU to CO" for the sample and referencegas respectively. The reference gas is carbondioxide prepared from Belemniiella americanafrom the Peedee formation of South Carolina.The slope of this curve agrees within the limitsof error with that secured by McCrea (1949;1950, p. 849) from inorganic crystals of calciumcarbonate grown at constant temperature. Thisequation is believed to be accurate to ±1°Cif the calcium carbonate were deposited in seawater of the mean isotopic composition of seawater of salinity 34.8%0. If the calcium car-bonate was deposited in water of other isotopiccomposition, appropriate corrections must bemade. The maximum variation in temperaturefor fully marine conditions due to the variationof isotopic composition of the oceans is ±4.5°C.

The variation of isotopic composition of thepresent oceans, as discussed by Epstein, Buchs-baum, Lowenstam, and Urey (1951), presentsthe greatest difficulty so far encountered in themeasurements of paleotemperatures. These au-thors point out that isotopic composition variesapproximately linearly with salinity. Their dataon surface waters are not very extensive, butthis rough linear relation applies also to surfacewaters. The isotopic composition varies in arough linear way with the temperature of sur-face waters. This indicates that, if no correctionfor the isotopic composition were made, a tem-perature of 30°C would be read as about 25°C,while a temperature of 10°C would be read as15°C. High temperatures would be read toolow, and low temperatures too high. However,no correction to past temperature determina-

tions can be made by use of their curve because(1) A general average change in the earth'stemperature would not change the averageisotopic composition of the oceans, and hencea measured isotopic composition could not becorrected with their curve. (2) The slope of thecurve of O18 composition vs. temperature is dueto evaporation and condensation of water, andhence can be expected to be greater in timesof high precipitation and lower in those oflow precipitation. In general one would expectthat high rates of evaporation and precipitationwould accompany higher average temperaturesof the earth's surface. However, we have thephenomena of glacial periods which somehowappear between limits of extremely low andextremely high temperatures, and these makeestimates in regard to the relation of tempera-ture to precipitation very difficult. For thepresent we do not attempt to estimate thecorrection for isotopic variations in the sea,but this problem is being studied from severalangles and it is hoped that corrections can beapplied in the future.

MEASUREMENTS OF ISOTOPIC COMPOSITION,AND TEMPERATURES OF THE

UPPER CRETACEOUS

In considering geological specimens for ourfirst attempts at measuring these temperatures,we have been guided by our considerations ofprobable diffusion of the carbonate in solidcalcium carbonate and the improbability thatrecrystallized carbonates would retain therecord. Since aragonite is unstable and com-monly recrystallized, all recrystallized arago-nitic shells will probably lose the record.

A study of fossil carbonate shells indicatedthat the belemnites would be the most likelyto retain the record. These have a very compactstructure with calcite crystals megascopicallyvisible and so closely fitted together that in-ternal reflections do not occur since the speci-mens are characteristically translucent. Thepolarizing microscope shows that the calcitecrystals are arranged with their axes perpendic-ular to the axis of the guard. The orientation

PLATE 1—SEASONAL GROWTH OF JURASSIC BELEMNITEFIGURE 1.—CROSS SECTION or A JURASSIC BELEMNITE

Showing relation of samples to growth rings. W and 5 refer to winter and summer regions respectively.FIGURE 2.—VARIATION or TEMPERATURE AND C13 CONTENT WITH RADIUS OF JURASSIC BELEMNITE

BULL. GEOL. SOC. AM., VOL. 61 URETrfo/ . , PL. 1

- --

oe i.oRADIUS

1.2

FK.UHE 2

ISOTOPIC^COMPOSmON AND TEMPERATURES OF UPPER CRETACEOUS 407

of the crystal axes and the shape and size of thecrystals are evident.

Several paleontologists advised investigationof the possibility that various representativespecies of the genus Inoceramus might retain

respect to carbon and oxygen. A flat diskabout 3 mm. thick was cut from the specimen,polished, and photographed in transmitted light(PI. 1, fig. 1). Samples of carbonate wereground from the specimen following the growth

TABLE 4.—ANALYSES or A JURASSIC BELEMNITE

Ring No.

12345678Q10111213141516171819, 202122, 23

atC")

R

3.3,3.573.822.652.4,3.274.2,3.522.66

3.463.29

L

2. S03.3S3.9,2.8,2.373.694.3,——2.763.623.4»

3.37—3.0, 3.083.1,

3.0s 3.1,2.762.742.592.6,

24 2.6S

Av.

3.4,3.882.7,2.3,3.4,4.3,3.5,2.652.763.5,3.343.37—3.073.1,3.1,2.752.7,2.5,2.6,2.66

«(0»)

R

-0.65-0.47-0.4,-1.0,-1.3,-1.37-1.2,-1.1,-1.0,—

-0.62-1.0,

L

-1.8,-0.9,-0.3o-0.8,-1.37-l.ls-1.1,——

-1.0s-0.4,-0.86

-0.8,—

-1.5, -1.3,-0.7,

-0.83 -0.6,-0.7.-1.2,-1.2,-0.8,-1.30

Av.

-0.6,(?)-0.7,-0.38-0.9,-1.3,-1.2,-LI?-l.U-1.0,-1.0,-0.5,-0.9,-0.8,—

-1.4,-0.7,-0.7,-0.7,-1.2,-1.2,-0.8,-1.3o

CC

15.,16.,14.,17.619.,19.,18.,18.,18.c18.,15.,17.417.,—20.316.,16.216.519.o19.217.,19.,

Av. 17.,

the temperature record. These fossils showevidence of massive calcite crystals of substan-tial size, though they are opaque. The calciticlayer of the shell is also present in many present-day mollusks, though usually not developed tothe remarkable extent that it is in this extinctgroup. Experiments in progress confirm ourexpectation that this group retains the temper-ature record.

In the belemnites, a particularly convincingtest for the durability of the record has beenmade. A Jurassic belemnite* approximately 2.5cm in diameter, brownish and translucent,with very well-defined growth rings, has beeninvestigated for a seasonal effect and bilateralsymmetry of isotopic constitution both with

4 Unfortunately, we do not know the species ofthis specimen. It was collected along with others onthe Island of Skye by Dr. Cyril S. Smith, who kindlymade them available to us. The specimens are ofvarious shapes and sizes and do not appear to be asingle species. As nearly as we can learn they areJurassic, probably from the Oxford clay.

lines from each side of the disk, as shown bythe numbered rings shown in the photograph.In some cases right and left samples were nottaken and analyzed, either because of accidentin the cutting or preparation or because thesamples were combined to secure sufficientcarbonate for our samples. In one case boththe right and left samples were lost.

The samples were analyzed for the C13 andO18 isotopes (Table 4).

The right and left samples of the outermostring do not agree either in C13 or O18, and thecheck for the two samples of the second ringis rather bad for the O18. Otherwise the agree-ment on the two sides lies within 0.2%0.Considering probable errors of measurement,and the difficulty of cutting samples of thesame thickness from the two sides, the agree-ment is very satisfactory. A radial variation ofisotopic composition occurs both in the carbonand oxygen isotopes. The variation of thecarbon isotopes does not correlate with any

408 UREY ET AL.~TEMPERATURES^OF UPPER CRETACEOUS

other variable found either in fossil or present-day specimens, and the variation in this speci-men is of the order found in present-dayspecimens. The erratic variation is probablydue to fractionation of the carbon isotopes byplants and the large effect that this has on thevery small amount of carbon in the sea (ap-proximately 28 mg. per liter).

Figure 2 of Plate 1 plots temperatures deducedfrom our O18 abundances against a radius meas»ured along a line from the center of growthoutward. The temperatures are plotted as hori-zontal lines over the range of the radius forwhich they are the mean temperature. Thetrue temperature curve should intercept equalareas above and below these lines. Such acurve is drawn trying to take account of thisrequirement and the uncertainty of the data.Where analyses for both the right and leftsides are available the line is the average, andpoints above and below represent the two data.The plot of S (Cla) is also given and shows nocorrelation with the seasons. The lack of struc-ture at the smaller radii is probably due to thefact that the samples are too thick to show thevariations that may exist.

It seems difficult to believe that any "acci-dental" process of diffusion or exchange couldhave produced such a distribution of carbonand oxygen isotopes, and we conclude that therecord has been substantially preserved for overa hundred million years by this belemnite. Inour work on other belemnites, we have neverhad any reason to doubt the isotopic recordproviding the specimen was translucent.

This Jurassic belemnite records three sum-mers and four winters after its youth, whichwas recorded by too small amounts of carbonatefor investigation by our present methods;warmer water in its youth than in its old age,death in the spring, and an age of about fouryears. The maximum seasonal variation intemperature is about 6°C. The mean tempera-ture was 17.6°C.

Dr. C. J. Stubblefield of the British Museumhas very kindly co-operated in securing speci-mens of belemnites, oysters, and brachiopodsfrom the Upper Cretaceous Chalk of England.Thus we have well-preserved specimens im-bedded in a matrix which has been modifiedbut slightly with time. Table 5 lists the speci-

mens secured from him and his record of theplaces and zones from which they originate.

Dr. J. Troelsen of the Mineralogical andGeological Museum of Copenhagen also sup-plied us with three belemnites from the lowerand upper Maestrichian of Denmark. The dataregarding their collection are as follows:

1. Belemnitella mucronata, Upper Maestri-chian, 17m. below the top of the whitechalk, Stevns, Denmark.

2. Belemnitella mucronata, Upper Maestri-chian, 30-40 mm. below the top of thewhite chalk, Stevns, Denmark.

3. Belemnitella, mucronata, Lower Maestri-chian (white chalk), M0en, Denmark.

In studying the isotopic composition of thesespecimens we first removed carefully the chalkfrom the specimen. In the case of the belemnitesthe surface layers and any chalky infiltrationsin cracks of the specimen were carefully re-moved. When possible a complete cross sectionof the guard was used to secure a true averagetemperature. In any case a piece of the solidclear calcite was gound to a fine powder andcarbon dioxide prepared from these samples.In the brachiopods and oysters the chalk againwas carefully removed, and parts of the cleanshells were gound to fine powder and usedfor the preparation of the carbon dioxide sam-ples. Whenever the chalk was available samplesof the chalk in the immediate neighborhoodof each specimen were analyzed. The resultsof the carbon and oxygen analyses for theEnglish samples are recorded in Tables 6 and7. In these tables the zones have been numberedas indicated in Table 5.

The oysters, brachipods, and their chalksagree in O18 composition only very roughly, andtheir C13 concentrations do not agree withinthe limits of our measurements. The belemnitesdo not agree in O18 concentrations with theirchalks nor with the oysters and brachiopods.We conclude that only the belemnites retainthe temperature record. Probably the crystalsof the other specimens are so small and thestructure so open that diffusion has destroyedthe record. In other studies now in progress wehave found that a chalky belemnite—i.e., onethat has lost its translucent character—agreesin isotopic composition with its surroundingchalk.

ISOTOPIC COMPOSITION AND TEMPERATURES OF UPPER CRETACEOUS 409

(-,o

§ 2sj;

nu on

1.SQ

§a|O o "3

TIs

H.B

o.

I

aliti

e

s

1

1

1

1et

8

\

•§ -2 art § § w^ fe w ^ jdg S ^ S^J _, ^ ^ ^

*C -S .3 53 <3 ^ QH"3 ^ S S S S•̂ 3 ."S "3 ^S *3 Cjs s *•* ^ G ^

CQ K) X -^ -^ -^

^ "1 ?JD <U _Qt- ^ M

|| JSligSlg| gjglg

"3 ;§• "3 ^ "3 ;§• "* "13 i g ^ g l § §

5 6 6 6 6 6 6 6

1̂!«•^ . r o ^ O ^ O O O O C N O O O

Hu_'>;

3 o2mO

£> ^a S•̂ s w

•2 S'S''•S. §•«

1 ! * si'i U ife § •rj! ^ 'S ^* S> ^

i l l 1 1 U 1 Sf* CS rr} ^O f^« ^o ON OTH

•CTrrTJTT'^ ^TIT.!*! O21 lVXlt_/ OL^LClClll

rH

13

oO00^

"S•g

Tere

brat

ulin

a lo

taIn

ocer

amus

labi

atus

3S3TVHOaia-aiH

§

a

ts

CO

C/2O

6

21"/

JA."

man

telli

ana

erby

)

Actin

ocam

axpl

enus

i !

niii

3TVHO

u«raou9g OBraomx | trenreraoaac)

410 UREY ET AL— TEMPERATURES OF UPPER CRETACEOUS

TABLE 6.—THE QUANTITY «(C13) FOR UPPERCRETACEOUS OF ENGLAND

1234i

6789

10

111213

Brachiopod

Speci-men

1.392.832.582.902.40

2.32——

1.5,1.9,

———

Chalk

0.941.8,1.6i—

1.80

1.6,

1.5,0.7s

1.17———

Oyster

Speci-men

0.792.60

1Q• ofl

2.2,1.6,

1.87

1.6v2.0,1.30

1.2,—

2.84

Chalk

0.7,1.5s1.7,1.7sl .U

1.6,~

I.St1.0,0.9«

1.22—2.8!

Belemnite

Speci-men

0.462.1,1.1.,0.9s—

1.85————

——

3.20

Chalk

1.641.3,1.3,——

___————

———

Some specimens of Table 5 were not used becauseof bad preservation.

constant composition had been reached, allthe analyses should be identical, and thus weconclude that this is not the case.6

The oxygen analyses for the Danish speci-mens together with derived temperatures aregiven in Table 8. The specimens are numberedas given previously.

Upper Cretaceous paleotemperature data forNorth America were secured from the Exogyracostata zone of a number of widely scatteredoutcrops of the Gulf Coast and Eastern Sea-board of the United States between westernTennessee and northeastern Mississippi north-eastward to New Jersey.' The specimens fromTennessee were derived from the sandy marlsof the basal third of the exposure on DaveWeeks' place, 3| miles south of Enville, Mc-Nair County, the type locality of the CoonCreek tongue of the Ripley formation (Wade,

TABLE 7.—THE QUANTITY 5(0") FOR UPPER CRETACEOUS OF ENGLAND

Zone

1234

Brachiopod

Specimen

-1.4,-1.9,,-2.40-1.96

5 -2.4,6 -2.1o

Chalk

-2.3,~2.9»-2.4,

Oyster

Specimen

-1.8s-2.6,-2.9,

- ! -1.9!-2.20 -2.7,-2.4,•7 :

8 —9 -2.8,

10 -2.6011 —12 —13 —14 —

-2.6,

-2 At

Chalk

-2.2,-2.9,-2.6,-2.6,-2.74-2.5,

— i —-2.89 -2.35

Belemnites

Specimen

-0.56-0.4,-1.24-1.2,

—-2.0,

——

-2.94 -2.0s -2.5, i —-2.4, -3.68 -2.7s- -3.62 , -3.1,— —— -1.89

!

—-3.5,

—

——

-0.40—

Chalk

-1.88-2.5s-3.85—

————

BelemniteTemperature

•cIS.o14.419.219.,

23.8

— ;—————

14.2

Present-day specimens show marked varia-tion in C13 concentrations even 'when they grewin the same region at the same time. Thus thesamples considered here should have had iden-tical or nearly identical O18 concentrations, butnot identical C13 concentrations when firstburied. If both the oxygen and carbon con-centrations changed with time we would expectthat the oxygen concentrations would tend tochange by comparable amounts, and henceremain somewhat the same as observed. On theother hand the carbon concentrations shouldshow no such approximate agreement. If com-plete equilibrium with water and carbon of

1926). Though belemnites do not occur in thesedeposits, depriving us of our reliable standard

5 We have analyzed a few samples of ooze fromthe Pacific and Atlantic Oceans in co-operation withW. F. Libby and J. R. Arnold, who are working onC14 concentrations in such materials. We are unableto interpret the results in any consistent way as yet,and evidently more data are required.6 The specimens from Western Tennessee, Mis-sissippi, North Carolina, and South Carolina werecollected by H. C. Urey and H. A. Lowenstam. Weare indebted to Dr. L. W. Stephenson for advisingus on certain collecting localities in South Carolina,to Dr. Horace C. Richards who guided us to theNew Jersey outcrops, and to Mr. P. A. Bethany ofMacon, Mississippi, who advised us on and collectedwith us from the outcrops of the Macon area.

ISOTOPIC COMPOSITION AND TEMPERATURES OF UPPER CRETACEOUS 411

for comparison, the excellent shell preservationof all the molluscan elements with originallyaragonitic shells, elsewhere in the outcrop areaspreserved only as casts, implied exceptionallyfavorable burial conditions suggestive of pres-ervation of original isotopic abundances. Thespecimens analyzed from northeastern Missi-ssippi were secured from the Prairie BluffChalk, at two localities, the glades on the VannyJackson farm west of the Rocky Hill Churchroad, 3.6 miles north-northeast of Starkville,Octibbeha County, which is the general areaof fossil locality 6846 of Stephenson and Monroe(1940), and from glades on the A. Lindley farm4 miles southwest of Macon, Noxubee County(Stephenson and Monroe's 1940 collecting lo-cality 17242, 17484). At both localities, thespecimens were found loose in the residualfossil concentrations which littered the glades,since Prairie Bluff specimens satisfactorily pre-served and located in situ were unavailable.The positions marked 10 feet and 13 feet abovethe base of the Prairie Bluff formation in Table9 for the specimens from the Lindley localityrefer to collecting levels on the glade and notto stratigraphic positions. Despite the richassemblages in the glade concentrations, onlyoriginally pure calcitic skeletal forms on thecalcitic layer of calcite and aragonitic inter-layered shells are preserved as such, reducingthe number of potential elements for paleo-temperature determinations to crinoidal re-mains, echinoids, Terebratulina floridana, Os-treidae, and Pectinidae among the pelecypodsand guards of Belemnitella americana, the bal-ance consisting of molds, prevalently in phos-phatic form. Shark teeth occur with the inver-tebrate assemblages composed of carbonateskeletons and furnish source material for apossible future check of the carbonate ther-mometer by the phosphate thermometer.

The South Carolina specimens were securedin situ from the basal section of the Peedeeformation which is well exposed in the westbank of the Peedee River at Burches Ferry,15 i miles from Florence. This is the localityfrom which Stephenson (1923) described andfigured a number of well-preserved specimensof Belemnitella americana and described a newspecies. At the time of our visit, the river waslow enough to expose a total of 13 J feet of Pee-

dee strata, consisting of calcareous, argillaceous,glauconitic sand above 3| feet of the shales ofthe Black Creek formation. Belemnitella ameri-cana was found in small numbers through theentire local Peedee section; it was concentratedin conspicuous numbers in a narrow 9-inchzone, 57 inches above the base of the formation.

TABLE 8.—OXYGEN ISOTOPIC ANALYSES ANDTEMPERATURES OP THE DANISH

BELEMNITES

Specimen

123

8(0") •

-0.17 !-0.43 }-0.0, |

re

12.814.312.3

The Belemnitella guards are throughout orientedmore or less horizontally, with occasional de-viations ranging up to about 10° from the hori-zontal plane. Within the horizontal plane thebelemnite rostra display random orientationas far as could be ascertained in the verticalexposure. Exogyra coslata occurs sparingly dis-tributed through the basal 7J feet of thesection, the individuals being generally smalland thin shelled. Larger, but comparativelyfragile individuals are abundant in the over-lying 22 inches of the section which is con-spicuously fossiliferous, consisting of a semi-coquina of pelecypod shell debris. The speci-mens analyzed were taken in situ.

The North Carolina specimens were collectedfrom a 6-inch calcareous sandstone ledge of thePeedee formation, exposed above water level,and 6 inches of soft marl below water level atthe Jackson's Hole Landing, locality of Ste-phenson (1923, p. 25), Northeast Cape FearRiver. A number of large, heavy-shelled Ex-ogyra costata individuals, with strongly resorbedsurfaces, and a single rostrum of Belemnitellaamericana, were secured in situ from the water-covered marly layer, which contrasts stronglywith the rich mollusk assemblages of overlyingcalcareous sandstone ledge. Pectinidae andBelemni'dla americana, originally calcitic skele-tal types alone are well preserved; the originallyaragonitic shells of the associated pelecypodsis often represented by casts and molds. As thePeedee section lacks top and bottom it is notpossible to determine the stratigraphic positionof this section within the formational range.

412 UREY ET AL—TEMPERATURES OF UPPER CRETACEOUS

TABLE 9.—OXYGEN ISOTOPIC ANALYSES AND TEMPERATURES FROM Bdemnildla amerkana FROM THESOUTHEASTERN UNITED STATES

Formation

Prairie Bluff

Peedee

Navesink

Locality

Starkville, Miss. 3.6 milesN-NE. on Rocky HillChurch Road

Macon, Miss. A. Lindleyplace 4 miles SW. ofMacon

Peedee River, S. C.

Cape Fear River

Mullica Hill, N. J.

Hornerstown, N. J.

Crosswicks Creek

Cream Ridge

Level

Unknown

13 feet above base of PrairieBluff

10 feet above base of PrairieBluff

105 inches above base offormation

57 inches to 68 inches abovebase of formation

Unknown

Base of formation

Base of formation

Unknown

Analysis

(1) -0.5,(2) -0.9,(3) -1.2,(4) -1.0,(5) -O.li

(1) O.Oo

(1) -1.0,(2) -0.7,(3) -0.5,(4) -1.4,

(1) -0.5:

(1) -0.44(2) -O-.So(3) -0.2,(4) -0.3,(5) -0.2,(6) -0.6,(7) -0.5,(8) -0.4,(9) -0.28

(10) -0.66(11) -0.6!

(1) -0.5«(2) -0.8,

(1) -1.2,(2) -1.1,(3) -0.90(4) -0.8,(5) -l.Oo(6) -0.9,

(1) -0.8,(2) -0.5,(3) -0.4,;4) -0.76(5) -0.8»(6) -0.8o(7) -0.3,(8) -0.6,

(I) -0.5,(2) -0.70

Temp.

H.,17.,19.,18.212.5

11.,

17.,16. j14.920.6

14.,

14.,13.613.413.,13.415.815.,14.,13. i15.,15.5

Av.Temp.

16.,

11..

17.,

14.,

14.,

15., j17., 16.,

19.,18.,17.216.,18.,17.,

17.oIS.o14.,16.,16.,16.,14. i15.,

15.416.o

18.o

16.o

15.,

ISOTOPIC COMPOSITION AND TEMPERATURES OF UPPER CRETACEOUS 413

The New Jersey specimens analyzed weresecured from the Navesink marl at the well-known fossil locality in the hillside just southof the railroad trestle at Mullica Hill (Weller,1907, locality 169), in the cut of a tributary of

Ul I

zUJ

t-fflUJ

Off

comxxa, and Belemnitella americana. The densepacking of the pelecypod shells with the ran-domly oriented, interwedged Belemnitellas in-dicates post mortem concentrations swept to-gether in shallow agitated waters. Burial

>•UJ

rlilUl

UlCOCCU

UlZ

ZQ

*!§ozpDI-UJUltOtO

COONCREEK

•?PRAIRIEBLUFF

RIPLEY

SELMACHALK

PEEDEE PEEDEE

EXOGYRACANCELLATA

MAESTRICHIAN

CAMPANIAN

NAVESINKMARL "

FIGURE 1.—CRETACEOUS SECTIONS FOR LOCALITIES STUDIED

Crosswicks Creek near Hornerstown, and fromthe dump piles of the abandoned "marl pit"on Edgar Schank's farm near Cream Ridge(Weller, 1907, locality 127). The Mullica Hillspecimens were derived from a 4-inch shell bed,consisting of Exogyra costata, Pectinidae, andBelemnitella americana in a marly glauconiticsandy matrix about 12 feet above the base ofthe bluff exposure. The preservation of the pele-cypods and belemnites is generally poor dueto leaching and partial limonitic replacement,while conspecific forms occurring in shell beds,located 10 feet lower and about 3 feet higherup in the bluflf section are represented only inthe form of casts and molds. The Hornerstownspecimens are derived from a shell bed, com-posed principally of Exogyra costata, Gryphaea

grounds foreign to the habitat are thus stronglyimplied for the Belemnitella element. Accordingto Spengler and Peterson (1950), the shell bedsmark the base of the Navesink member of theMonmouth formation, previously interpretedas a formation and group respectively. A basalposition for the Mullica Hill and Hornerstownspecimens in the Navesink is thus indicated,while the stratigraphic position of the CreamRidge forms, secured loose from the dump piles,cannot be determined.

All the North American specimens includedin the present study fall within the range of theExogyra costata zone, as shown by the correla-tion of the formations from which they werederived (Fig. 1). In comparing the stratigraphicpositions of the specimens secured from Tennes-

414 UREY ET A L.—TEMPERATURES OF UPPER CRETACEOUS

see, Mississippi, and South Carolina, and theMullica Hill and Hornerstown localities in NewJersey within their respective formations andin relation to the formational ranges and theircorrelation, it will be seen that the Coon Creekforms, from southwestern Tennessee, occupy alower; those from the Peedee formation of SouthCarolina and from the Navesink localities ofNew Jersey an intermediate, and the PrairieBluff forms from northeastern Mississippi ahigher position within the zonal range. Theirstratigraphic relations to the North Carolinaand Cream Ridge locality of New Jersey cannotbe determined because of the unknown strati-graphic positions of the latter within their re-spective formations. To judge from the narrowrange of the Navesink marl, the New Jerseyspecimens may be anywhere between the top ofthe Coon Creek formation of southwestern Ten-nessee and the base of the Prairie Bluff chalkin northeastern Mississippi, those from the Pee-dee formation of North Carolina a considerablywider interval.

The North American forms serve for an ini-tial survey of the special paleotemperaturerange within the time interval occupied by theExogyra costata zone, even though in some casesour specimens came from different levels withinthat zone and in other cases the relationshipswithin the zone are uncertain.

DISCUSSION OF THE NORTH AMERICANTEMPERATURES

The Exogyra costata, zone is correlated withthe Maestrichian of the European standard sec-tion (Stephenson and Reeside, 1938, p. 1637;Stephenson et al., 1942, Chart No. 9). Belem-nitella mucronatz which occupies in the UpperSenonian of northern and central Europe astratigraphic position similar to Bekmnitellaamericana, in southeastern North America, hasbeen considered closely related to the NorthAmerican species (Stephenson, 1923, p. 56).Approximate time equivalency of their enclos-ing strata is further indicated by other faunaland floral similarities. Keeping in mind the as-yet existing uncertainties in intercontinentalcorrelation, particularly with reference to de-tails in time equivalency and length of timeintervals, it is nevertheless permissible to com-pare in general terms the Maestrichian paleo-

temperature data from both sides of the Atlan-tic that is, of the Gulf Coastal area in NorthAmerica and of southern England and Den-mark in northwestern Europe. There is sur-prisingly little difference in the temperaturesrecorded by the reliable forms from either sideof the Atlantic despite the latitudinal range ofthe collecting localities, which extend from 33°to about 41°N. in North America to betweenabout 52° and 56°N. in England and Denmark.

In surveying the meaning of the temperaturerecords only those of the belemnites can be atpresent considered reliable. As the temperaturesrecorded constitute at best mean temperaturesof skeletal growth periods, they do not neces-sarily represent annual mean temperatures ofthe waters in which they lived. This is clearlyindicated by the temperature range recordedby the shell of a recent Chama macerophyllafrom the Bermuda inshore waters, which hasbeen analyzed for seasonal effects. Shell ma-terial is laid down only after the surface watersreach 24°C and extends through the warm partof the year, which reaches a maximum at 29°C,while the yearly temperature range of the in-shore waters covers a range from 18°C to 29°C.The mean temperature of 26J°C recorded by theChama mazerophylla shell is thus 2J°C higherthan that of the waters. Organisms with anarrow temperature range for skeletal growth,particularly with a growth period largely off theaverage with reference to the yearly range oftheir waters, would constitute unreliable datafor climatic records unless forms from identicalhabitat with wider or overlapping skeletalgrowth periods are available for comparison.The Inoceramus lineage now under investiga-tion may furnish a comparative element forcorrection at a future date.

Returning to the belemnites, preference forcooler waters is strongly implied by their geo-graphic distribution. Dacque (1915, p. 423-426), in discussing the paleoclimatology of theCretaceous, pointed out that broad climaticzoning is introduced in the beginning of theCretaceous and becomes sharply establishedby Upper Cretaceous time. Differentiation intoa boreal and Mediterranean-aquaetorial zonesis fairly well documented, the aquaetorial zonebeing characterized by prominent carbonatesediments, reef-forming corals and rudistids,

THE NORTH AMERICAN TEMPERATURES 415

large Foraminifera, and heavy-shelled Nerine-idae and actaeonellids. Most convincing is theappearance of dwarfed and crippled rudistidsin the bordering boreal zones both to the northand south, as shown by Dacqu£'s distributionmap for the rudistids. Belemnitella was shownto be confined to the northern boreal zone ofthe Upper Cretaceous seas. In comparing thesize ranges of the collections of Belemnitellaamericana secured from the various localities ofthe Gulf Coastal area, an increase in maximumsizes attained is clearly noticeable from Missis-sippi northeastward to New Jersey. The Missis-sippi localities were apparently located in thesouthern border area, as Stephenson and Mon-roe record representatives of the rudistid genusDurania from the Exogyra cancellata subzoneof the upper Selma Chalk in which Belemnitellaamericana makes its first appearance.

The northeastward trend of increasing sizesin Belemnitella americana could be correlatedwith cooler water preference, but their meantemperature records do not seem to supportthis view. Refinement in the sampling procedurecontemplated for studies of seasonal variationsmay decide this question. Throughout our dis-cussion we have neglected the possibility andeven probability of some variation of the iso-topic composition of the waters, and this mayexplain the nearly identical temperatures of theCarolina and Mississippi region as contrastedto New Jersey.

The depth range is another factor yet to beconsidered in the evaluation of the tempera-tures recorded by nectonic elements such asbelemnites. The sediment character and theburial assemblages of the American representa-tives in the locations examined indicate that,though the bottoms were largely below effectivewave base (except for New Jersey examples,which occur in wave concentrated shell beds),the abundant and varied mollusk forms pointto comparatively shallow bottoms, well withinthe euphotic zone. This agrees with Naef's(1922) conclusions of prevalent shallow-waterhabitat for the belemnites in general as impliedby functional morphology. Thus the tempera-ture variation should be characteristic of alimited depth range. That Belemnitella lived inshallow cooler waters is further indicated by theabsence of stray individuals in the southernTethys waters.

The narrow gradient indicated by the meantemperatures of the Belemnitellas across thelatitudinal width from Mississippi to Denmarkis surprising. Unless the temperature toleranceof skeletal growth of the European and Ameri-can species happens to coincide, rather uniformclimatic conditions are suggested over most ofthe boreal zone during the Maestrichian. Grad-ual lowering of the temperatures through UpperCretaceous time toward the Maestrichian areindicated in England and agree with the estab-lishing of the climatic zones noted by Dacque.Increased mixing of the oceans due to strongercurrent development should have resulted.North America and South America were proba-bly separated by the circumequatorial Tethys,and the waters from the pole area previouslyconnected with it via the western interiorstrait became separated at about the criticaltime that the lowering of the late Cretaceoustemperatures is indicated to have taken place.Free mixing of the northern waters due to lackof the present existing north-south land barriersshould have tended to equalize the tempera-tures toward the pole. Subtropical climaticconditions could have thus been maintained onthe northern land masses where the dinosaurscontinued to flourish.

Cowles (1939; 1940; 1945) has suggested ex-tinction of the dinosaurs at the closing intervalof the Cretaceous due to increased tempera-tures. On the other hand continued lowering ofthe temperatures as suggested by the Englishdownward trend would have equally causedtheir extinction. We do not wish to emphasizethis apparent downward trend since our tem-peratures depend on single samples in each zoneand our experience indicates that a variationbetween a number of specimens of the samelocality of as much as 5°C is possible. The in-vestigation of more specimens from these zonesand seasonal variations of such specimens mayenable us to reach a definite conclusion. Longi-tudinal paleotemperature traverses for the en-tire Upper Cretaceous extending into earlyTertiary, based on a large number of associatedfaunal elements and including benthonic formsshould furnish a clearer trend of paleoclimato-logic conditions as well as aid in quantitativelyevaluating the temperature niches of the marineorganisms.

416 UREY ET AL.—TEMPERATURES OF UPPER CRETACEOUS

REFERENCES CITED

Colbert, E. H., Cowles, R. B., and Bogert, C. M.(1946) Temperature tolerances in the Americanalligator and their bearing on the habits, evolution,and extinction of the dinosaurs, Am. Mus. Nat.Hist., Bull., vol. 86, p. 333-373.

Cowles, R. B. (1939) Possible implications of rep-tilian thermal tolerance, Science, vol. 90, p. 465-466.(1940) Additional implications of reptilian sen-

sitivity to high temperature, Am. Nat., vol. 75,p. 542-561.

(1945) Heat-induced sterility and its possiblebearing on evolution, Am. Nat., vol. 79, p. 160-175.

Curie, P. (1889) Recherches sur le Pouvoif inducteurspecifique et le conductibUitf des corps cristallises,Ann. Chim. Phys., vol. 17, p. 385.

Dacqufi, E. (1915) Grundlagen und Methoden derPalaeogeographie, Gustav Fischer, Jena.

Dole, M. (1935) Relative atomic weights of oxygen inwater and air, Am. Chem. Soc., Jour., vol. 57,p. 2731.

—— (1936) Relative atomic weights of oxygen inwater and air. Discussion of the atmospheric dis-tribution of the oxygen isotopes and of the chemicalstandard of atomic weights, Chem. Phys., Jour.,vol. 4, p. 268.(1949) The history of oxygen, Science, vol. 109,

p. 77-81.Epstein, S., Buchsbaum, R., Lowenstam, H. A.,

and Urey, H. C. (1951) Carbonate-water isotopictemperature scale, Geol. Soc. Am., Bull., vol. 62,p. 417-425.

Goldschmidt, V. M. (1933) Grundlagen der quanti-tativen Geochemie, Fortschritte Mineral., Kris-tal., Petrog. vol. 17, p. 112.

Joffe, A. (1923) ElekirizitiUsdurchgang durch Krys-tatte, Ann. Phys., vol. 72, p. 461.

McCrea, J. M. (1949; 1950) Isotopic chemistry ofcarbonates and paleotemperature, Doctoral Dis-sertation, Univ. Chicago; Chem. Phys., Jour.,vol. 18, p. 849 (1950).

Murphey, B. F., and Nier, A. 0. C. (1941) Varia-tions in the relative abundance of the carbonisotopes, Phys. Rev., vol. 59, p. 771.

Naef, A. (1922) Die fossilen Tintenfische, GustavFischer, Jena, p. 322.

Nier, A. O. C., and Gulbranson, E. A. (1939) Vari-ations in the relative abundance of the carbon iso-topes. Am. Chem. Soc., Jour., vol. 61, p. 697.

Spengler, W. B., and Peterson, J. J. (1950) Geologyof Atlantic Coastal Plain in New Jersey, Dela-ware, Maryland and Virginia, Am. Assoc. Pet-rol. Geol., Bull., vol. 34, p. 40.

Stephenson, L. W. (1923) The Cretaceous formationsof North Carolina, N. C. Geol. Econ. Survey,vol. V, pt. I.and Monroe, W. H. (1940) Upper Cretaceous

deposits, Miss. Geol. Survey, Bull 40,296 pages.Reeside, J. B., Jr. (1938) Comparison of Upper

Cretaceous deposits of Gulf region and WesternInterior region, Am. Assoc. Petrol. Geol., Bull.,vol. 22, p. 1629-1638.et al. (1942) Correlation of the outcropping Cre-taceous formations of the Atlantic and GulfCoastal Plain and Trans-Pecos Texas, Geol. Soc.Am., Bull., vol. 53, p. 435-448.

Thode, H. G., MacNamara, J., and Collins, C. B.(1949) Natural variations in the isolopic contentof sulfur, and their significance, Canad. Jour.Research, vol. 27, p. 361., , Lossing, F. P., and Collins, C. B. (1948)Natural variations in the isotopic content of boronand its chemical atomic weight, Am. Chem. Soc.,Jour., vol. 70, p. 3008.

Urey, H. C. (1947) The thermodynamic properties ofisotopic substances, Jour. Chem. Soc., p. 562.and Greiff, L. J. (1935) Isotopic exchange equi-

libria, Am. Chem. Soc., Jour., vol. 57, p. 321.Wade, B. (1926) The fauna of the Ripley formation

on Coon Creek, Tenn., U. S. Geol. Survey, Prof.Paper 137.

Weller, S. (1907) Cretaceous paleontology of NewJersey, Geol. Survey N. J., vol. IV, Paleont.Ser.

INSTITUTE or NUCLEAR STUDIES, UNIVERSITY orCHICAGO, CHICAGO, ILL.

MANUSCRIPT RECEIVED BY THE SECRETARY OF THESOCIETY, MARCH 7, 1950.

PROJECT GRANT 517/47/S48.

ADDENDUM

It is necessary to consider the possible effect of the changed temperature scale mentioned inthe Addendum of the preceding paper on the temperatures recorded here. Repetition of the analysesusing the modified method of preparation and the corrected temperature scale on 10 specimensreported in this paper changed their measured temperatures by an average of 1.1 °C with a spreadof —0.6° to +2.2°C. The temperatures remain remarkably uniform though systematically higherby about 1°C.