JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 97, NO. B10, PAGES 13,917-13,951, SEPTEMBER 10, 1992 A New GeomagneticPolarity Time Scale for the Late Cretaceous and Cenozoic STEVENC. CANDE1AND DENNISV. KENT Lamont-DohertyGeological Observatoryof Columbia University,Palisades, New York We have constructeda magneticpolarity time scalefor the Late Cretaceousand Cenozoicbased on an analysisof marinemagnetic profiles from the world's oceanbasins. This is the first time, sinceHeirtzler et al. (1968) publishedtheir time scale,that the relative widths of the magneticpolarity intervalsfor the entire Late Cretaceousand Cenozoichave been systematicallydetermined from magneticprofiles. A compositegeomagnetic polarity sequencewas derivedbased primarily on data from the South Atlantic. Anomaly spacingsin the South Atlantic were constrainedby a combinationof finite rotationpoles and averagesof stackedprofiles. Fine-scaleinformation was derived from magnetic profiles on faster spreadingridges in the Pacificand Indian Oceansand insertedinto the SouthAriantic sequence. Based on the assumptionthat spreadingrates in the South Atlantic were smoothly varying but not necessarily constant,a time scalewas generatedby using a splinefunction to fit a set of nine age calibrationpoints plusthe zero-ageridge axis to the co•npositepolarity sequence. The derivedspreading history of the South Atlantic showsa regular variation in spreadingrate, decreasingin the Late Cretaceousfrom a high of almost70 mm/yr (full rate) at aroundanomaly 33-34 time to a low of about30 mm/yr by anomaly27 time in the early Paleocene,increasing to about 55 mm/yr by about anomaly 15 time in the late Eocene,and then graduallydecreasing over the Oligoceneand the Neogeneto the recentrate of about 32 mm/yr. The new time scalehas severalsignificant differences from previoustime scales. For example,chron C5n is -0.5 m.y. older and chronsC9 throughC24 are 2-3 m.y. youngerthan in the chronologiesof Berggrenet al. (1985b) and Hadandet al. (1990). Additionalsmall-scale anomalies (tiny wiggles)that representeither very shortpolarity intervalsor intensityfluctuations of the dipole field have been identifiedfrom several intervals in the Cenozoic including a large number of tiny wiggles between anomalies 24 and 27. Spreadingrates on severalother ridges, including the SoutheastIndian Ridge, the East PacificRise, the Pacific-AntarcticRidge, the Chile Ridge, the North Pacific,and the CentralAtlantic, were analyzedin order to evaluatethe accuracyof the new time scale. Globally synchronousvariations in spreadingrate that were previouslyobserved around anomalies 20, 6C, andin the late Neogenehave been eliminated. The new time scalehelps to resolveevents at the timesof •najorplate reorganizations.For example,anomaly 3A (5.6 Ma) is now seento be a time of suddenspreading rate changesin the SoutheastIndian, Pacific-Antarctic, and Chile ridgesand may correspondto the ti•ne of the changein Pacificabsolute plate motionproposed by others. Spreading rates in the North Pacific became increasingly irregular in the Oligocene, culminatingin a precipitousdrop at ano,naly6C time. INTRODUCTION because the HDHPL68 time scale was constructed based on the The geomagnetic polarity time scale of Heirtzler et al. assumptionthat the spreading rate in the South Atlantic has [1968] (HDHPL68; see Table 1 for acronymsof time scalesused been constant. Because of the fundamental importance of the in this paper) was one of the foundationsof the plate tectonic geomagnetic polarity time scale to numerous disciplines, we revolution. Building on the seafloor spreading hypothesis of sought to improve the resolution of the relative polarity widths Vine and Matthews [1963] and the remarkablesymmetry found by analyzing magnetic anomaly profiles now available from by Pitman and Heirtzler [1966] in the Eltanin 19 magnetic throughoutthe world's oceanbasins. anomalyprofile acrossthe Pacific-AntarcticRidge, Heirtzler et al. [1968] proposeda geomagneticpolarity time scale for the BACKGROUND Late Cretaceousto Recent, about the last 75 Ma (from anomaly 32), based on a few long magnetic anomaly profiles. The Heirtzler et al. [1968] constructed their time scale by successof HDHPL68 can be gaugedby the fact that although assuminga constant spreadingrate for a profile collected on parts of the reversalsequence have been subsequentlyrevised, R/V Vema, cruise 20, in the South Atlantic. They then nearly half of the original relative spacings of reversals incorporateda polarity sequencederived from the higher spread- (nominally anomalies6C to 29) have been used in virtually all ing rate North Pacific according to the curve for the relative revisionsto the present[e.g., Harland et al., 1990]. spreading rate between the two oceans. A time scale It has also becomeincreasingly clear that there are errorsin extrapolatedfrom the South Atlantic profile was selected as a the relative widths of the polarity intervals as given in standardbecause (1) the anomaly pattern for the South Indian HDHPL68 and in the subsequenttime scalesthat are basedon it. Ocean was not sufficiently long, (2) the North Pacific profile For example, Barker [1979] used HDHPL68 to analyze a was too distortednear the ridge axis for extrapolation,and (3) magnetic profile collected along a flow line in the South the spreadinghistory in the South Pacific was variable and gave Atlantic and found that there were several rapid changes in an extrapolatedtime scale that becameunacceptably too young. spreadingrates over the Cenozoic. The discovery of sudden The evolution of the geomagneticpolarity time scale since changesin spreadingrate in the South Atlantic is suspicious 1968 involved two types of revision: adjustmentof the relative ,, spacingof some anomaliesgiven by HDHPL68 and calibration 'Now at ScrippsInstitution of Oceanography,La Jolla, California. of the polarity sequencein time. Most of the revisions to the relative spacingswere done over relatively short segmentsof Copyright1992 by the AmericanGeophysical Union. HDHPL68. Talwani et al. [1971] proposedrevisions to the relative ages of anomalies2 to 4A based on a detailed study of Papernumber 9ZIB01202. the Reykjanes Ridge. Larson and Pitman [1972] added 0148-0227/92/9ZIB-01202505.00 anomalies33 and 34 to the sequence. Blakely [1974] stacked 13,917 1•,918 CANDEAND •: NEW GEO•GNETIC POLARITY• SCM• TABLE 1. GeomagneticPolarity Time ScalesReferred to in A more accuraterepresentation of spreadinghistory can be This Paper obtainedby determiningclosely spacedfinite rotation poles basedon magneticanomalies and fracturezone trendsalong an entire spreadingridge. Finite rotationpoles take into account Time Scale Reference asymmetricspreading, ridge jumps and other local spreading irregularitiesthat can distort the magneticanomaly sequence along a particularprofile. For example,Cande et al. [1988] HDHPL68 Heirtzler et al. [1968] calculateda set of 43 finite poles that constrainthe spreading historyof the SouthAtlantic for the last84 m.y., fromanomaly LKC77 LaBrecqueet al. [1977] 34 to the ridge axis. Representativedistances between anomaliescan be determinedalong a syntheticflow line based LA81 Lowrie and Alvarez [1981] on the finite rotation poles. BKFV85 Berggren eta/. [1985b] In Figure1 we comparethe spreading rate history of theSouth Atlantic for three different time scales(HDHPL68, GTS89, and GTS89 GeologicTime Scale 1989 Berggrenet al. [1985b] (BKFV85, whichis the sameas Kent (Harland eta/., 1990) and Gradstein [1986]), referencedto the distancesto anomalies along a syntheticflow line derived from the finite rotation CK92 Cande and Kent (this paper) polesof Cande et al. [1988]. We note that there are large variationsin apparentspreading rate even with HDHPL68 (Figure 1, bottom),which in fact assumeda constantrate of several profiles from the northeastPacific and revised the spreadingin the SouthAtlantic. A very similar patternof relative width of anomalies4A to 6. Klitgord et al. [1975] variation is seen between anomalies 6 to 29 in more recent time examined deep-tow magnetic profiles acrossseveral ridge scales(e.g., Figure 1, BKFV85 and GTS89), eventhough more segmentsof the Pacificand revised anomalies 1 to 3A. Cande age calibrationdata are used. While the spreadingrate and Kristoffersen[1977] revisedthe width of anomalies30 to variationsshown in Figure1 couldindicate very erraticrelative 34 from a global.analysis. platemotion, a moreplausible interpretation is that the single LaBrecqueet al. [1977] incorporatedthese changes in a profile (Vema 20) availableto and usedby Heirtzler et al. revised time scale (LKC77); the only modificationmade to the [1968], andincorporated in virtuallyall subsequenttime scales, entirePaleogene section of the sequencewas to deleteanomaly is simplynot representativeof the spreadinghistory of the 14 which was generally recognized to be an artifact in South Atlantic. It follows that observations of globally HDHPL68. However, LaBrecque et al. [1977] encountered difficultiesin splicingthe revisedportions of the sequenceinto 34 HDHPL68. For example,Blakely [1974] tied his revisionsof the relative width of anomalies4A to 6 to the youngerend of TS89 anomaly5 of HDHPL68. Sincethe relativewidth of anomaly 4A to 5 in Blakely [1974] is considerablylarger than in 3o 6,c •3 HDHPL68 [and Talwani et al., 1971], andno changewas made to the ageof anomaly3A as givenby Klitgordet al. [1975],the l / anomaly3A to 4A intervalis compressedin LKC77. This problemwas recognizedbut left unresolvedeven in the most recenttime scale(GTS89) of Harland