Geological Society of America Bulletin

doi: 10.1130/0016-7606(1951)62[417:CITS]2.0.CO;2 1951;62;417-426Geological Society of America Bulletin

HAROLD C UREYSAMUEL EPSTEIN, RALPH BUCHSBAUM, HEINZ LOWENSTAM and CARBONATE-WATER ISOTOPIC TEMPERATURE SCALE

Email alerting servicese-mail alerts when new articles cite this article

to receive freewww.gsapubs.org/cgi/alertsclick

Subscribe

Bulletinsubscribe to Geological Society of America

towww.gsapubs.org/subscriptions/click

Permission request

to contact GSAhttp://www.geosociety.org/pubs/copyrt.htm#gsaclick

positions of the Society.viewpoint. Opinions presented in this publication do not reflect official worldwide, regardless of their race, citizenship, gender, religion, or politicalforums for the presentation of diverse opinions and positions by scientists includes a reference to the article's full citation. GSA provides this and otherarticles on their own or their organization's Web site providing the posting posted to any Web site, but authors may post the abstracts only of theiruse in classrooms to further education and science. This file may not be and to make unlimited copies of items in GSA's journals for noncommercialfigure, a single table, and/or a brief paragraph of text in subsequent works granted permission, without fees or further requests to GSA, to use a singleemployees within scope of their employment. Individual scientists are hereby Copyright not claimed on content prepared wholly by U.S. government

Notes

the scope of their employment.by U.S. government employees within not claimed on any material preparedSociety of America, Inc. Copyright is Copyright © 1951, The Geological

on May 27, 2010gsabulletin.gsapubs.orgDownloaded from

BULLETIN OF THE GEOLOGICAL SOCIETY OF AMERICAVOL. 62. PP. 417-426. 6 FIGS. APRIL 1961

CARBONATE-WATER ISOTOPIC TEMPERATURE SCALE

BY SAMUEL EPSTEIN, RAIJH BUCHSBAUM, HEINZ LOWENSTAM, AND HABOLD C. UREY

ABSTRACT

A relationship between temperature and relative O18 abundance in calcium carbonate in marine shellshas been determined. If the relative O18 abundance of the water in which the shell grew is known, the tem-perature can be determined to an accuracy of ±1.0°C. The relative O18 contents of marine waters increasewith salinity and, in the case of surface marine waters, with salinity and temperature.

CONTENTSPage

Introduction 417Acknowledgments 417Relative O18 contents of marine waters 418Isotopic temperature scale 420

Marine shells selected and their growthtemperatures 420

Extraction of carbon dioxide from marineshells, the problem of impurities, andresults 422

Discussion 424Conclusions 425References cited 425Addendum 425

ILLUSTRATIONSFigure Page1. Plot of salinity vs. O18 content for ocean

water 4192. Plot of salinity vs. O18 content for surface

ocean waters 4193. Plot of temperature vs. O18 content for sur-

face ocean waters 4194. Plot of O18 concentration vs. temperature:

calcium carbonate samples without heattreatment 423

5. Plot of Ou concentration vs. temperature;calcium carbonate samples with heattreatment 424

INTRODUCTION

Urey et al. (1951) discuss a method formeasuring paleotemperatures that depends on avariation in the oxygen —18 abundances in cal-cium carbonate deposited in the sea. McCrea(1949; 1950, p. 849) investigated the isotopiccomposition of calcium carbonate precipitatedinorganically from water and determined thetemperature coefficient for the abundance ofO18 in calcium carbonate relative to water as afunction of temperature; he found a value of-5.4°C permil increase in the ratio of O18 toO16 in the calcium carbonate. McCrea gave acalculation for this quantity including estimatesof the effect of the translational and hinderedrotational degrees of freedom in the solid stateon the distribution functions, and arrived at avalue of 5.1. Assuming free rotation in the solidstate Urey (1947, p. 562) calculated a value of—5.7 for the same quantity. This research aimsto determine this quantity and to set up an

equation relating temperature to the differencein isotopic composition of samples of carbondioxide prepared from organically precipitatedcalcium carbonate and an arbitrarily selectedreference sample of carbon dioxide.

The relationships between temperature andO" abundance in calcium carbonate from ma-rine shells and that from inorganic calcium car-bonate precipitates are similar. But a numberof special considerations must be taken intoaccount in the case of marine shells to find thetemperature of the sea water in which the ma-rine animals lived: (1) the isotopic compositionof the oxygen of the oceans; (2) the collectionof marine shell-bearing animals which grow inwater of constant and known temperatures; and(3) the problem of impurities in the carbondioxide prepared from marine shells.

AKNOWLEDGMENTSWe are indebted to The Geological Society

of America and to the American Petroleum In-417

on May 27, 2010gsabulletin.gsapubs.orgDownloaded from

418 EPSTEIN ET AL.—ISOTOPIC TEMPERATURE SCALE

stitute for grants which have made this workpossible, and we wish to extend our most sincerethanks for this help. During recent months thework has also been supported under Task OrderXXVIII, Contract N6ori-20 of the Office ofNaval Research.

RELATIVE O18 CONTENTS OF MARINE WATERS

Since the ratios of vapor pressures of H2O"to H2O18 are 1.008 and 1.011 at 25°C and 0°Crespectively (Riesenfeld and Chang, 1936, p.127), evaporation and condensation continu-ously change the O18 contents of bodies of water.The magnitude of the variation of O18 in theoceans is important to the determination of avalid temperature scale and to the general prob-lem of the determination of temperature of un-known samples, because this thermometer de-pends essentially on the magnitude of thedifference between the O1* contents of the cal-cium carbonate and the water in which it wasprecipitated. Consequently a series of O18 deter-minations were made on samples of sea water,most of which were made available to us by H.Pettersson, who collected them on the ALBA-TROSS expedition (1949, p. 182). The method ofdetermining the relative O18 content of watersamples involves the equilibration of carbondioxide with the water at 25°C and analysis ofan aliquot of the gas. The carbon dioxide soequilibrated is compared with our standardworking sample of carbon dioxide prepared fromthe calcium carbonate of belemnites of the Pee-dee formation using a high-precision mass spec-trometer (McKinney, McCrea, Epstein, Allen,and Urey, 1950). The analyses are relativeonly, but are correct within a probable error ofabout 0.2%0 of the isotope ratio. The results ofthese analyses are given in Table 1. The unitused in this table is defined by the followingequation :

TABLE 1.—RELATIVE O18 CONTENT OF MARINEWATERS

S(0'8) = (|̂ - A\Kllanilard /1000,

where RSamPu and Retard are the ratios ofC016O18 to CO" for the sample and standardreference gas, respectively.

Apparently, the isotopic composition of oxy-gen is not homogeneous throughout the oceans.Deviations from the mean isotopic composition

Location

30°47'N. 25°25'W.

26°25'N. 34°00'W.14°27'N. 66°13'W.14°15'N. 71°02'W.02°52'S. 88°20'W.

13°25'S. 149°30'W.

07°45'S. 120°30'E.10°06'N. 52°45'E.

35°41'N. 21°SO'E.

00°13'S. 18°26'W.

28°05'N. 60°49'W.

43°04'N. 19°40'W.President Channel,

Puget SoundMiddle Channel,

Puget SoundEncenada, MexicoSt. Tomas, MexicoMonterey Bay,

CaliforniaLa Jolla, CaliforniaBermuda

Depthof pointof col-lection(Me-ters)

10100050005000

020

5001000

010004000

1020

5002000

502000

25500

25007500

25500

20002000

171

157

~0~0~0

~0~0

Tem-per-ature<°C)

———4.2

27.718.87.54.5

27.74.31.4

29.529.310.72.9

16.513.622.67.13.01.4

28.516.33.73.69.4

7.4

191511

1518-29

Salinity(%„)

37.4435.4334.934.936.034.434.634.635.834.634.734.535.535.034.838.838.735.634.734.934.8

s(O»)(%.)

-O.G!-0.5,-0.2!-0.74

0.42-0.3,-O.So-O.S6-0.2,-0.9,-0.66

0.570.0,

-0.2,-0.90

0.690.840.04

-0.50-0.3,-0.44

36.8 0.3836.3 ! -0.1835.035.031.4

33.1

-0.5,-0.67-1.94

-0.9,

33.5 -0.8!33 | -0.7733.8 -0.85

34.0 i -0.7,— ; 0.64

of the oceans occur in the waters close to themouth of a river, as is the case for the PresidentChannel location. If water of fully marine com-position is considered, then the maximum vari-ation of O18 in ocean water is 1.7%0, whichwould represent about 10°C variation in tem-perature determination. Clearly, it is necessaryto determine the O18 content of the water inwhich a marine shell grew if this particular shellis to be used in correlation of temperature withthe O18 content of organically precipitated shell.

on May 27, 2010gsabulletin.gsapubs.orgDownloaded from

RELATIVE O18 CONTENTS OF MARINE WATERS 419

On further examination of the data an at-tempt was made to establish a correlation be-tween the O18 content and the salinity of thewater. Such a correlation is possible becauseprocesses which tend to increase or decrease theO18 of the water also tend to increase or decrease

t 36Z

-10 -08 -0.6 -0.4 -O2 0.2 0.4 0.6 OS

FIGURE 1. — PLOT OF SALINITY vs. O18 CONTENT FOROCEAN WATERS

the salinity. The plot of salinity against relativeO18 content is shown in Figure 1. The pointsscatter badly, but a regular trend is evident,and a straight-line relationship is assumed be-tween salinity and O18 content of the water. Therelationship is only approximate as was to beexpected. Extrapolation of the line to zero salin-ity gives —13.5%0 as the relative O18 contentof fresh water.

The approximate linear relationship betweensalinity and O18 abundances suggests that itmight be more rigorously followed if limited tosurface waters. Since surface salinity and sur-face temperature show some indication of beingrelated linearly, we have plotted in Figures 2and 3 salinity vs. O18 content and temperaturevs. O18 content respectively for the waters ofTable 1 coming from a depth of 25 meters orless. The plots indicate rather definite linearrelations, but the data are too meager to justifyany definite conclusion. We shall continue thisinvestigation as we secure more water samples,particularly from higher latitudes.

The mean isotopic composition of the oceanswill be close to that of the water of mean sa-linity, and we have selected this as having—0.47%0 relative to our standard gas and

38

37

36

3*(0

34

33-I

oFIGURE 2.—PLOT OF SALINITY vs. O18 CONTENT FOR

SURFACE OCEAN WATERS

30

°C 20

10

FIGURE 3.—PLOT OF TEMPERATURE vs. O18 CON-TENT FOR SURFACE OCEAN WATERS

corresponding to 34.8%0 salinity. This gives—13.0%0 as the isotopic composition of freshwater relative to that of sea water of 34.8%0mean salinity. This difference is larger than thatexpected for a fractionation equal to the ratioof vapor pressures, and substantially greater thanthat for any fresh-water samples investigatedby us. Lake Michigan is representative and

on May 27, 2010gsabulletin.gsapubs.orgDownloaded from

420 EPSTEIN ET AL— ISOTOPIC TEMPERATURE SCALE

analyzes as — 6.2%0 relative to our mean seawater.

This large variation of isotopic compositionof sea water decreases the reliability of tempera-ture measurements by this method unless asample of water is analyzed at the same timeand a correction for the isotopic composition ofthe water is made on the carbon dioxide analy-ses. This error, some 2.5 times as large as ex-pected, is disappointing, especially when paleo-temperatures are considered. The melting iceof high latitudes must have a lower O18 contentthan has fresh water of the temperate zone. Thewhole problem of water evaporation from theoceans and its precipitation over the land andoceans is very complex, and our data are notsufficiently extensive to justify a more detaileddiscussion.

A correlation between salinity and Ols con-tent is useful in that salinity data which arequite readily available for present-day oceanscan give an approximate correction for the iso-topic abundance if water samples are not avail-able for mass spectrometric analyses. We hopethe distribution of 0" between phosphate andwater will give another thermometer with a dif-ferent temperature coefficient, thus makingpossible the elimination of the water phase.1

ISOTOPIC TEMPERATURE SCALE

Marine Shells Selected and their GrowthTemperatures

One problem in establishing a temperaturescale using marine calcareous-shelled animalswas to get shells which grew at known tempera-tures. To grow certain marine animals undercontrolled temperatures in the laboratory is alengthy and difficult project. Nevertheless, con-stant-temperature baths were set up at threePacific Coast marine laboratories,1 and these

1 M. Steinberg is developing this possibility inour laboratory.2 Oceanographic Laboratory, University of Wash-ington, Friday Harbor, Washington. We are in-debted to Dr. Thomas G. Thompson and Dr. EmerySwan for assisting in this work, and also to Dr. C.Barnes for help in locating suitable areas for collect-ing. At the Hopkins Marine Station we are indebtedto Dr. Rolf Bolin for help in looking after one of theexperimental tanks. At Scripps Institution of Ocean-ograph, La Jolla, California, we are indebted to Dr.Carl Hubbs who assisted us in setting up one of theexperimental tanks and in making collections inlower California.

were stocked with calcareous organisms withnotched or drilled shells. It was thought that ifrepair occurred, the new shell, when analyzed,would yield reliable data which could be usedto fix points on the temperature scale. Not allof the material is in, and some of the attemptswere unsuccessful. However one of the experi-mental batches of animals has been analyzedand the results are presented later. Con-sequently, shell-bearing marine animals werecollected from locations which have a minimumannual variation of temperature. The speciesof animals chosen were attached or sedentaryforms which live in relatively confined areasso that the point at which collection was madewould be representative of the condition underwhich the animals grew their shells. Animalswere collected from marine locations whichhad records of maximum constancy and whichhad a minimum of contamination with fresh-water drainage. Collections were made in thecooler waters of Puget Sound, intermediateconditions of Monterey Bay, and the wannerwaters of certain areas along the coast of lowerCalifornia in Mexico. In addition, shells fromthe still warmer waters of Bermuda were an-alyzed.

Particular care was taken to estimate care-fully the temperature at which the particularfragment of shell used for analyses grew. Forexample, the samples analyzed from SantaTomas, Mexico, included one large red abalone,Haliotis rufescens, and one large black abalone,Haliotis cracherodii, and two keyhole limpets,Fissuretta volcano. The Haliotis were collectedAugust 25, 1949, and thus the outer few milli-meters of the rim of the shell grew within themonths May to August. Dr. C. L. Hubbsmeasured temperatures at several stations inthis region during September 1948, May, July,and August 1949, and also observed the dailyvariation on August 18, 1949. From thesemeasurements we conclude that 15°C is areasonable average temperature. The variationsin temperature during these months is aboutthe same as the daily variation from 14.4°Cto 17.0°C. Thus a probable error of ±1°Cseems a reasonable estimate. A complete crosssection of the shell of the keyhole limpets wastaken so that it was necessary to estimate theaverage annual temperature. The temperaturein this location goes as low as 11°C and as high

on May 27, 2010gsabulletin.gsapubs.orgDownloaded from

ISOTOPIC TEMPERATURE SCALE 421

TABLE 2.—DESCRIPTION OP SHELL SAMPLES USED FOR DETERMINATION OF TEMPERATURE SCALE

Location

Middle Channel, PugetSound 48°22'30"N.,122°58'45*'W 130-157meters, July 22, 1949.

President Channel, PugetSound 48°41'50"N.,122°59'30"W 171 meters July 16, 1949

Hopkins Marine Station,Pacific Grove, Calif.

Santa Tomas, Lower Cali-fornia, Mexico Aug. 27,

Between Encenada andMorro Points, LowerCalifornia, Mexico. 3meters, Aug. 25, 1949

Bermuda, Tank in labo-ratory

No.

1

2

3

4

5

6

7

8

910

11

12

13

SpecimenName

Mussel Volsella modiolus

Brachiopod Terebratulinatransversa

Mussel Volsella modiolus

Brachiopod Terebratulinatransversa

Red abalone Haliotisrufescens. At low tidelevel May 14, 1949

Giant keyhole limpetMegathura crenalata 50m. off-shore, 3 m.Aug. 15, 1949

Red abalone Haliotisrufescens

Black abalone Halioliscracherodii

Limpets Fissurella volcano

Red abalone Haliotisrufescens

Snail (conch) Strombusgigas

Bivalve, Pinna sp.

Part of specimen used ^

Cross section of completehalf shell

Complete half shell

Cross section of completehalf shell

Cross section of completehalf shell

At least a year's growthof shell

Cross section of completeshell

Last J cm of large shell

Last J cm of large shell

Cross section of 2 com-plete shells

Edge of shell

Regenerated shell

Regenerated shell (part)

Est.Temp.°C

7.4

9.4

12.6

11.5

15

15

14

20

31

29

Water

-0.9,

-1.94

-0.8S

-0.77

-0.8,

0.6«

as 17°C. The average temperature of thissample of the shell was thus estimated to be14°C within an error of less than 3°C andperhaps as close as ±1°C. Similar considera-tions of temperature data made available to usled to the assignment of other temperatures,and similar uncertainties exist in the estimate.

Table 2 describes the location at which themarine animals were found, the estimated tem-perature of the fragment of shell used for analy-

ses, and the relative O18 of the water collectedwith the sample.

The carbon dioxide was prepared from themarine shells by reacting the calcium carbonatewith approximately 100 per cent ortho phos-phoric acid in an evacuated vessel. The carbondioxide was trapped in a liquid nitrogen cooledtrap and pumped free of the more volatileimpurities, then passed through traps covered

on May 27, 2010gsabulletin.gsapubs.orgDownloaded from

422 EPSTEIN ET AL— ISOTOPIC TEMPERATURE SCALE

by a solid COs-acetone bath to freeze out suchless volatile impurities as water.

The samples of the shells described in Table2 were then cleaned of obvious impurities and

isotopes of the calcium carbonate during theirdeposition in their complex shell structure.But not enough information has been obtainedto justify this conclusion regarding these groups

TABLE 3.—ISOTOPIC ANALYSES OF ORGANIC CALCIUM CARBONATE SAMPLESEach isotopic analysis is a mean of at least four sample preparations. The measurements were all within

the accuracy of ±0.2%0.

Number of spec-imen (Table 2)

123456789

1011

ObservedTemp. "C

7.47.49.49.4

12.611.51515141420

12 3113 29

Analysis (Nobeat treatment)

1.4,1.3,0.54

-0.2,0.501.0,0.920.960.600.460.27

~0.58—

Analysis (Withheat reatment)

0.360.1,

-1.2T-1.1«-O.Si-0.3i-0.97-0.8i-0.4,-O.So-1.80-2.27-1.7,

Water correction

0.440.441.4,1.4,0.3,0.3,0.300.3o0.300.300.34

-l.h-l.li

Corrected analysis(No heattreatment)

1.871.782.0!1.2,0.881.4,

Corrected analysis(With heattreatment)

0.8o0.570.200.3,

-0.1,0.0,

1.22 ! -0.671.250.9»

-0.5,-0.1,

0.76 -0.2»0.6i

-1.6,—

-1.4,-3.3,-2.8,

the carbon dioxide prepared. Table 3 gives theresults of analyses when this method of prep-aration of carbon dioxide is used. The speci-men numbers correspond to those in Table 2.The results of the sixth column are plotted inFigure 4.

The data show a correlation between tem-perature at which the marine animals grewand the relative abundance of O18 in the carbondioxide prepared from the shells. But the devia-tions from a linear or nearly linear relationshipare much larger than the expected experimentalerror. The difference in analyses for shells col-lected simultaneously from the same locationeliminated the possibility that the poor correla-tion between temperature and relative O" con-tent was due to incorrect estimation of thetemperature at which the animals grew.

The possibility that the animals did not laydown their shells in equilibrium with the sur-rounding water seemed to be unlikely becausethe relative O18 contents of calcium carbonateshowed a definite correlation with temperature.Some preliminary results obtained during theinvestigation indicate that perhaps the echi-noderms and corals fractionate the oxygen

of invertebrates. In any case, the shells hereused in the determination of the temperaturescale were those which are deposited externalto the animals' soft parts.

Comparison of results of McCrea with thoseof Table 3 indicated that the possible errors inthe present work were of the order of magnitudeof 1%0 of the mass 46 to mass 44 ratio. Sinceimpurities of mass 46 present to the extentof three parts per million would introduce anerror equivalent to 1%0 in the ratio, the usualchemical methods for analysis of impuritieswould be impractical in this case. Therefore,the criterion for valid analyses became the con-sistent correlation between temperature atwhich the shells were laid down and the meas-ured change in relative abundance of mass 46in the carbon dioxide.

Extraction of Carbon Dioxide, Problems ofImpurities, and Results

The method of collecting the ion currentsin the mass spectrometer gave measurementsof the abundance ratio of mass 46 to the sumof masses 43, 44, 45, and 47. Consequently,

on May 27, 2010gsabulletin.gsapubs.orgDownloaded from

ISOTOPIC TEMPERATURE SCALE 423

although the ratio of mass 46 to mass 44 wouldbe relatively insensitive to small concentrationsof impurities of masses 43, 45, and 47, thesehad to be considered. Purification of samples

30

ao-

SAMPLES WITHOUTHEAT TREATMENT

-30 -20 -10«(o°e)%=

FIGURE 4.—PLOT or O18 CONCENTRATION vs. TEM-PERATURE: CALCIUM CARBONATE SAMPLES

WITHOUT HEAT TREATMENT

could then either decrease or increase thevalue of the 46 to 44 ratio depending on whetherpurification eliminated impurities of mass 46or of masses 43, 44, 45, and 47.

Many molecules or radicals formed in theionizing chamber of the mass spectrometer,both organic and inorganic can contribute to theions of mass 43 to 47. Some of these can bedftOH, CS, CHaCOOH, BC1, NO2, N2O, andpropane. The first attempts were focussed onthe possible impurities in the carbon dioxidegas. Several carbon dioxide samples from differ-ent marine shells were mixed with small volumesof bromine gas and irradiated by strong light,with the hope that bromination would shiftthe mass range of the impurities beyond the43 to 47 range. Some samples of carbon dioxidewere mixed with Van Slyke's (1940, p. 509)oxidizing solution or passed over hot CuO.Organic impurities would then be convertedto carbon dioxide, and if present in traceamounts should not affect the 46 to 44 ratioof the carbon dioxide sample as a whole. Sam-ples were also mixed with traces of ammoniaand water to hydrolize compounds like boron

trichloride, and passed over hot copper toreduce possible traces of nitrogen oxides.

In all these cases no appreciable effect on theisotopic analyses was observed.

Samples of carbon dioxide from the shellswere also mixed with fluorine. Only wherethere was evidence of incomplete separationof the carbon dioxide and fluorine and sub-sequent contact with glass was there a changeinO18 analysis. These changes were erratic forany one source of gas samples and varied from-0.3%o to -200%o of the 46 to 44 ratio. Thiswas obviously due to SiF+ of mass 47 addingto ion beam 44 and thus showing a decrease inthe ratio of 46 to 44.

A more successful approach to the purifica-tion problem involved treatment of the pow-dered calcium carbonate before preparation ofthe carbon dioxide. The powder was heated at400°C for 60 minutes with a slow stream ofhelium passing over the powder. The heliumwas first purified by passing it over copper oxideat 750°C and then through a liquid nitrogencooled trap. With this treatment the relativeO18 contents of the calcium carbonate fromshells was decreased in most cases, and in areproducible manner.

Experiments to determine the length of timenecessary for heating the calcium carbonatepowder to attain constancy of O18 analysis withfurther heating showed that heating at 400°Cfor 15-30 minutes was necessary to "purify"the samples tried. A safety factor of two wasadopted, and the samples were heated for anhour. Since the equilibrium pressure of carbondioxide over calcium carbonate is 10~6 atmos-pheres at 400°C, it was necessary to test thistreatment for possible loss of carbon dioxideand fractionation of the oxygen isotopes. Twosamples of calcium carbonate, one normal inO18 and the other enriched by 100%0, wereprecipitated inorganically (using sodium car-bonate and calcium chloride). Samples of eachwere analyzed without the heat treatment andwith heat treatment. The analyses of eachvariety of calcium carbonate checked, showingthat heat treatment did not change the analysesof such samples. Clearly, it was the organicallyprecipitated calcium carbonate alone whichwas affected by the heat treatment. To shortenthe time for the experiments the powdered

on May 27, 2010gsabulletin.gsapubs.orgDownloaded from

424 EPSTEIN ET AL.—ISOTOPIC TEMPERATURE SCALE

samples were heated for 15 minutes at 475°Cwith the attainment of the identical resultsobserved for the longer heating times. We donot understand the mechanism of purification

masses 46 and 44 for the sample and standardgas. The constant 11.88 is characteristic of ourstandard gas, while the constant 5.91 shouldbe independent of this gas.

35

30

20

15

10

SAMPLES WITHHEAT TREATMENT

-4 - 3 - 2 - 1 0 I 25(0")%.

FIGURE 5.—PLOT OF O18 CONCENTRATIONS vs.TEMPERATURE: CALCIUM CARBONATE SAMPLES ...3

WITH HEAT TREATMENT

of the marine shell samples by the heating proc-ess. In most cases it is obvious that charringof organic matter occurs.

Calcium carbonate samples from specimenslisted in Table 2 were heated before the carbondioxide was prepared, and then analyzed. Theresults are given in Table 3, last column, andplotted in Figure 5.

Comparisons of Figures 4 and 5 indicatethat the heat treatment of the organicallyprecipitated calcium carbonate lowered the ap-parent 01! contents of the carbon dioxide pre-pared from all these samples. Apparently themarine shells contained variable amounts ofsome impurity, causing the inconsistency inthe relationship between temperature andOu

content when unheated samples were used.Least-squares calculations of the data for

the heat-treated samples in Table 3 gave astraight-line relationship

( = 11.88 - 5.9la(O18),

where t equals the temperature in °C, S(C)18) isthe per mil difference between the ratios of

DISCUSSIONOur constant, 5.91, for the slope of the curve

differs from that secured by McCrea (5.4). Thedeviation is in the direction expected for in-complete equilibrium between solid calciumcarbonate and dissolved carbonate ion as foundby him. We do not believe this is the cause ofthe disagreement since a considerable varietyof organisms all give analyses lying so closelyon our curve.3 The measurements of these smallquantities are very difficult and, though we donot understand the reasons for the disagree-ment, it is only 10 per cent in the slope of thecurve. Rather than regarding this as disagree-ment, we believe it shows that these living or-ganisms approximate equilibrium conditionsrather closely. Analyses at present are morereproducible than they were when McCreacompleted his work, and this may have signifi-cance in regard to the discrepancy.

The possible effects of the difference in crys-tal structure of calcite and aragonite have notbeen explored. It is hardly to be expected thatany large effects will be present, but, as ourtechniques improve, they should be investi-gated. Not only may the isotopic compositionof aragonite and calcite crystals grown undersimilar conditions differ, but our heating pro-cedure may convert aragonite to calcite andthus modify fractionation effects in the liber-ation of carbon dioxide by phosphoric acid.

The presence of a physiological effect in thecase of certain groups of animals such as theechmoderms and corals, and plants such as acoralline algae, has seemed probable at somepoints in our work, but we do not feel it iscertainly established. These specimens are diffi-cult to investigate because of the large amountsof organic matter present, but, since we haveestablished a rather secure basis for our workon the mollusks and brachiopods, it now seemsfeasible to investigate these groups. The echino-

3 As this is written an analysis of a sample ofregenerated shell from Haliotis crafherodii grown ina bath at a controlled temperature of 18.9 ± 0.1°C.shows agreement with our curve within 0.3°C.

on May 27, 2010gsabulletin.gsapubs.orgDownloaded from

DISCUSSION 425

derms and corals are not particularly usefulin relation to the paleotemperature scale be-cause of extensive infiltration of the fossils. TheForaminifera should be investigated for, evenif not useful over long periods of time, theymay be of interest for the investigation of morerecent temperatures.

The variation of Ou content in marine andfresh waters, together with precise methods ofmeasurement of the small quantities involved,makes possible an investigation of meteorologi-cal effects of considerable interest. The fraction-ation of the oxygen isotopes in evaporation andprecipitation of water should give some infor-mation in regard to these processes as theyoccur in nature. We intend to continue ourstudies on the isotopic composition of watersince the problem bears so directly on themeasurements of paleotemperatures.

CONCLUSIONS

The variation in isotopic composition ofmarine waters is appreciable in terms of theisotopic temperature scale. When correctionsare made for the isotopic composition of seawater in which mollusks and brachiopods havelived, a linear relationship between temperatureand the Ou concentration exists. The slope ofthe curve agrees approximately with calcu-lated values and with that secured when in-organically precipitated calcium carbonate sam-ples crystallized at known temperatures are

used. Organically precipitated calcium carbon-ate gives reproducible analyses if it is firstheated to 475°C for 15 minutes and thenreacted with phosphoric acid to prepare carbondioxide, the gas used in the mass spectrometer.

REFERENCES CITEDMcCrea, J. M. (1949; 1950) Isolopic chemistry of

carbonates and paleotemperature, Doctoral Dis-sertation, Univ. Chicago; Chem. Phys., Tour,vol. 18, p. 849 (1950).

McKinney, C. R., McCrea, J. M., Epstein, S.,Allen, H. A., and Urey, H. C. (1950) Im-provements in mass spectrometers for the measure-ment of small differences in isotope abundanceratios, Rev. Sci. Inst., vol. 21, p. 724.

Pettersson, H. (1949) The floor of the ocean, Endeav-or, vol. 8, p. 182.

Riesenfeld, E. H., and Chang, T. L. (1936) Vaporpressure, boiling point, and heat of vaporizationof EDO and #2018, Z. Phys. Chem., vol. B33,p. 127.

Urey, H. C. (1947) The thermodynamic properties ofisotopic substances, Jour. Chem. Soc., April, p.562., Lowenstan, H. A., Epstein, S., and McKin-ney, C. R. (1951) Measurement of paleotemper-atures and temperatures of the Upper Cretaceousof England, Denmark, and the SoutheasternUnited States, Geol. Soc. Am., Bull., vol. 62,p. 399-426.

Van Slyke, D. D., Folch, J., and Plazin, J. (1940)Manometric carbon determination, Biol. Chem.,Jour., vol. 136, p. 509.

INSTITUTE TOR NUCLEAR STUDIES, UNIVERSITY orCHICAGO, CHICAGO, ILLINOIS.

MANUSCRIPT RECEIVED BY THE SECRETARY OF THESOCIETY, MARCH 8, 1950.

PROJECT GRANT 517/47/S48.

ADDENDUM

We have delayed publication of this paper while it was in galley proof because we discovereddisturbing variations in our temperature determination. We traced this to the use of heated copperoxide to remove impurities from helium gas passed over our samples during the heating process.Small amounts of oxygen from the copper oxide were present in the helium, and the carbon dioxideformed apparently exchanged in some cases with the calcium carbonate of the samples. Removalof all oxygen by means of a charcoal trap at liquid nitrogen temperature corrected these difficulties.Our tentative corrected temperature equation becomes < = 14.8-5.415. The slope now agrees withMcCrea's value. We will publish details of this temperature scale shortly.

on May 27, 2010gsabulletin.gsapubs.orgDownloaded from