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 Cenozoicbasedon an

analysisof marinemagneticprofilesfrom the world's oceanbasins. This is the first time, sinceHeirtzler
et al. (1968) publishedtheir time scale,that the relative widthsof the magneticpolarity intervalsfor the
entire Late Cretaceousand Cenozoichave been systematicallydeterminedfrom magneticprofiles. A
compositegeomagneticpolarity sequencewas derivedbasedprimarily on data from the SouthAtlantic.
Anomaly spacingsin the South Atlantic were constrainedby a combinationof finite rotationpoles and
averagesof stackedprofiles. Fine-scaleinformation was derived from magnetic profiles on faster
spreadingridgesin the PacificandIndian Oceansandinsertedinto the SouthArianticsequence.Basedon
the assumptionthat spreadingrates in the South Atlantic were smoothlyvarying but not necessarily
constant,a time scalewas generatedby usinga splinefunctionto fit a set of nine age calibrationpoints
plusthe zero-ageridgeaxis to the conposite
polaritysequence.The derivedspreading
historyof the South
Atlantic showsa regular variation in spreadingrate, decreasingin the Late Cretaceousfrom a high of
almost70 mm/yr (full rate) at aroundanomaly33-34 time to a low of about30 mm/yr by anomaly27 time
in the early Paleocene,increasingto about55 mm/yr by about anomaly 15 time in the late Eocene,and
then graduallydecreasingover the Oligoceneand the Neogeneto the recentrate of about32 mm/yr. The
new time scalehas severalsignificantdifferencesfrom previoustime scales. For example,chronC5n 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-scaleanomalies(tiny wiggles)that representeither
very shortpolarity intervalsor intensityfluctuationsof the dipolefield have been identifiedfrom several
intervals in the Cenozoic including a large number of tiny wiggles between anomalies 24 and 27.

Spreadingrateson severalotherridges,includingthe Southeast

IndianRidge,the EastPacificRise, the
Pacific-AntarcticRidge,the Chile Ridge, the North Pacific,andthe CentralAtlantic,wereanalyzedin order
to evaluatethe accuracyof the new time scale. Globally synchronous
variationsin spreadingrate that were
previouslyobserved
aroundanomalies
20, 6C, andin thelate Neogenehavebeeneliminated.The new time
scalehelpsto resolveeventsat the timesof najorplate reorganizations.For example,anomaly3A (5.6
Ma) is now seento be a time of suddenspreadingrate changesin the SoutheastIndian, Pacific-Antarctic,
and Chile ridgesandmay correspondto the tine of the changein Pacificabsoluteplate motionproposed
by others. Spreading rates in the North Pacific became increasinglyirregular in the Oligocene,
culminatingin a precipitousdrop at ano,naly6C time.
INTRODUCTION

because the HDHPL68

The geomagnetic polarity time scale of Heirtzler et al.

[1968] (HDHPL68; seeTable 1 for acronymsof time scalesused
in this paper) was one of the foundationsof the plate tectonic
revolution. Building on the seafloor spreading hypothesisof
Vine and Matthews [1963] and the remarkablesymmetryfound
by Pitman and Heirtzler [1966] in the Eltanin 19 magnetic
anomalyprofile acrossthe Pacific-AntarcticRidge, Heirtzler et
al. [1968] proposeda geomagneticpolarity time scale for the
Late Cretaceousto Recent,aboutthe last 75 Ma (from anomaly
32), based on a few long magnetic anomaly profiles. The
successof HDHPL68 can be gaugedby the fact that although
partsof the reversalsequencehave been subsequently
revised,
nearly half of the original relative spacings of reversals
(nominally anomalies6C to 29) have been used in virtually all
revisionsto the present[e.g., Harland et al., 1990].
It has also becomeincreasinglyclear that there are errorsin
the relative widths of the polarity intervals as given in
HDHPL68 and in the subsequenttime scalesthat are basedon it.
For example, Barker [1979] used HDHPL68 to analyze a
magnetic profile collected along a flow line in the South
Atlantic and found that there were several rapid changes in
spreadingrates over the Cenozoic. The discovery of sudden
changesin spreadingrate in the South Atlantic is suspicious
,,

'Now at ScrippsInstitutionof Oceanography,

La Jolla, California.
Copyright1992 by the AmericanGeophysicalUnion.

Papernumber9ZIB01202.
0148-0227/92/9ZIB-01202505.00

time scale was constructed

based on the

assumptionthat the spreading rate in the South Atlantic has

been constant. Because of the fundamental importanceof the
geomagnetic polarity time scale to numerous disciplines, we
sought to improve the resolutionof the relative polarity widths
by analyzing magnetic anomaly profiles now available from
throughoutthe world's oceanbasins.
BACKGROUND

Heirtzler et al. [1968] constructed their time scale by

assuminga constantspreadingrate for a profile collected on
R/V Vema, cruise 20, in the South Atlantic. They then
incorporateda polarity sequencederived from the higher spreading rate North Pacific according to the curve for the relative
spreading rate between the two oceans. A time scale
extrapolatedfrom the South Atlantic profile was selected as a
standardbecause(1) the anomaly pattern for the South Indian
Ocean was not sufficiently long, (2) the North Pacific profile
was too distortednear the ridge axis for extrapolation,and (3)
the spreadinghistory in the South Pacific was variable and gave
an extrapolatedtime scale that becameunacceptablytoo young.
The evolution of the geomagneticpolarity time scale since
1968 involved two types of revision: adjustmentof the relative
spacingof some anomaliesgiven by HDHPL68 and calibration
of the polarity sequencein time. Most of the revisions to the
relative spacingswere done over relatively short segmentsof
HDHPL68. Talwani et al. [1971] proposedrevisions to the
relative ages of anomalies2 to 4A basedon a detailed study of
the Reykjanes Ridge. Larson and Pitman [1972] added
anomalies33 and 34 to the sequence.Blakely [1974] stacked

13,917

1,918

CANDEAND:

NEW GEOGNETIC POLARITY

TABLE 1. Geomagnetic
PolarityTime ScalesReferredto in
This Paper
Time Scale

Reference

HDHPL68

Heirtzler

LKC77

LaBrecqueet al. [1977]

LA81

Lowrie and Alvarez [1981]

BKFV85

Berggren eta/. [1985b]

GTS89

CK92

et al. [1968]

SCM

A more accuraterepresentation
of spreadinghistorycan be
obtainedby determiningclosely spacedfinite rotationpoles
basedon magneticanomaliesandfracturezonetrendsalongan
entire spreadingridge. Finite rotationpolestake into account
asymmetricspreading,ridge jumpsand otherlocal spreading
irregularitiesthat can distortthe magneticanomalysequence
alonga particularprofile. For example,Candeet al. [1988]
calculateda set of 43 finite poles that constrainthe spreading
historyof the SouthAtlanticfor thelast84 m.y., fromanomaly
34 to the ridge axis. Representativedistancesbetween
anomaliescan be determinedalong a syntheticflow line based
on the finite rotation poles.

In Figure1 we compare
thespreading
ratehistoryof theSouth
Atlantic for three different time scales(HDHPL68, GTS89, and

GeologicTime Scale 1989

Berggrenet al. [1985b](BKFV85, whichis the sameas Kent

(Harland eta/., 1990)

and Gradstein [1986]), referencedto the distancesto anomalies

Cande and Kent (this paper)

several profiles from the northeastPacific and revised the

relative width of anomalies4A to 6. Klitgord et al. [1975]
examined deep-tow magnetic profiles acrossseveral ridge

segments
of the Pacificandrevisedanomalies
1 to 3A. Cande
and Kristoffersen[1977] revisedthe widthof anomalies
30 to

along a syntheticflow line derived from the finite rotation

polesof Cande et al. [1988]. We note that there are large
variationsin apparentspreadingrate even with HDHPL68
(Figure 1, bottom),whichin fact assumeda constantrate of
spreadingin the SouthAtlantic. A very similarpatternof
variation is seen between anomalies 6 to 29 in more recent time

scales(e.g., Figure1, BKFV85 andGTS89), eventhoughmore

age calibrationdata are used. While the spreadingrate
variationsshownin Figure1 couldindicatevery erraticrelative

platemotion,a moreplausibleinterpretation
is thatthe single
profile (Vema 20) availableto andusedby Heirtzler et al.
[1968], andincorporated
in virtuallyall subsequent
time scales,
revisedtime scale (LKC77); the only modificationmade to the
is
simply
not
representative
of
the
spreading
history
of the
entirePaleogene
sectionof the sequence
wasto deleteanomaly
South Atlantic. It follows that observations of globally
14 which was generally recognized to be an artifact in
34 from a global.analysis.

LaBrecqueet al. [1977] incorporated

thesechangesin a

HDHPL68.

However, LaBrecque et al. [1977] encountered

difficultiesin splicingthe revisedportionsof the sequence

into
HDHPL68. For example,Blakely [1974] tied his revisionsof

34

the relative width of anomalies4A to 6 to the youngerend of

anomaly5 of HDHPL68. Sincethe relativewidthof anomaly

4A to 5 in Blakely [1974] is considerablylarger than in

TS89

3o

HDHPL68 [andTalwani et al., 1971], andno changewasmadeto

the ageof anomaly3A asgivenby Klitgordet al. [1975],the

anomaly3A to 4A intervalis compressed
in LKC77. This
problemwas recognized
but left unresolved
evenin the most

6,c

recenttime scale(GTS89) of Harland et al. [1990].

Age calibrationhas beenthe focusof changesto the

geomagneticpolarity time scale since the compilationof
LaBrecque
et al. [1977]. HDHPL68assumed
a constant
rateof
seafloorspreadingin the.SouthAtlantic, and useda single
calibrationpointat the olderend of anomaly2A to extrapolate
agesout to anomaly32. LaBrecqueet al. [1977] addeda
calibrationpoint at the older end of anomaly29, basedon
magnetobiostratigraphic
data which placedthe Cretaceous/
Paleogene
boundarywithinthe uppermost
partof chronC29r
[Lowrie and Alvarez, 1977]. This left the assumptionof
constant spreadingin the South Atlantic largely intact.
However,subsequent
calibrationefforts addedmore control

pointsbasedon developing
magnetobiostratigraphic
ties, and
relaxed the implicit assumptionof constantspreadingin the

20

BKFV85

20
' 5?
$2

2O

SouthAtlantic to smallertime intervals[e.g., Nesset al., 1980;

Lowrie and Alvarez, 1981;Harland et al., 1982, 1990;Berggren
et al., 1985b;Haq et al., 1988].
ACCURACYOFTHE GEOMETRICPOLARITYTIME SCALE

One way to evaluatethe accuracy

of a geomagnetic
polarity
time scaleis to calculatethe consequences
on spreadingrateson
0
20
40
60
80
variousridges. Ideally,a "type"profilewouldbe availablefor
AGE (Ma)
everyridgesystem
thatcouldbe usedto determine
thespreading
ratehistoryof thatridge. In reality,no singlemagnetic
profile Fig. 1. Comparison
of half spreading
ratesin the SouthAtlanticusing
reflectsthe completespreading
historyof a ridgesystemdueto various time scales,as identified in the text. The distancebetween
is calculated
alonga synthetic
flowlineat 30 S usinga setof
irregularities
suchas periodsof asymmetrical
spreading,
small anomalies

ridgejumpsandpropagating
rifts.

43 rotationpolesfromCandeet al. [1988].

CANDE AND KENT: NEW GEOMAGNETICPOLARtrY TIME SCALE

13,919

REFERENCE SECTION FOR THE SOUTH ATLANTIC:

synchronousspreading variation [e.g., Vogt, 1986] based on
CATEGORY I AND II DISTANCES
HDHPL68 or derivativetime scalesshouldbe regardedas suspect
andmay well be artifactsof errorsin the polarity sequence.This
Cande et al. [1988] describedthe spreadinghistory of the
problem motivates our reassessment of the geomagnetic
South Atlantic since anomaly 34 with a set of 43 finite rotation
polarity time scale in terms of the relative width of polarity
poles. We used nine of these rotation poles to establish a
intervalsas a framework for age calibration.
framework for averagingthe widths of anomalieson individual
CONSTRUCTING A NEW TIME SCALE
profiles. The choice of nine rotation poles was somewhat
Constructionof a magnetic anomaly time scale involves age arbitrary, and, in fact, we could have used a larger subset.
However, the analysisof shorterand larger numbersof profiles
calibration of a reference sequence of relative anomaly
becomesprogressivelymore unwieldy as more poles are used.
spacings. Our initial approach for a reference section was to
Distances to key anomalies were constrained along a
search for long magnetic profiles from different oceans that
individually were high fidelity recordsof the polarity reversal synthetic flow line in the South Atlantic at (half separation)
pattern over 10 to 20 m.y. time intervals. This procedurewas intervals of 150 to 300 km (roughly 7 to 13 m.y.) based on
first suggestedby Klitgord and Schouten[1986] and later by finite rotation poles for the younger endsof anomalies4A, 5C,
Aubry et al. [1988], who compared pairs of profiles from 7, 13, 20, 24, 30, 33, and 34 [Cande et al., 1988] (Table 2).
These category I intervals were chosen, in part, so that
different ocean basins and identified sections several hundred
kms long that appearedto have formed at a constantspreading anomalies with ambiguous reversal boundaries, such as the
rate. We found, however, that it was extremely difficult to youngend of anomaly5 which has fine-scalestructure,were not
assemblea referencesectionand devise a schemefor calibrating at the end of an interval. The synthetic flow line, which is
located at approximately30S, and the location of the distance
it for the entire Late Cretaceous and Cenozoic, based on
intervals constrainedby the finite rotation poles are shown in
individual profiles from different ridges. Individual long
profiles, as we noted earlier, almostinvariably include disconti- Figure 2.
nuities due to ridge jumps, propagating rifts, intervals of
asymmetricspreadingor small fracture zones; even when such TABLE 2. Intervals of Normal Polarity Crustal Magnetization
Along a Synthetic Flow Line in the South Atlantic
interruptionsin continuity are identified, residual uncertainties,
especiallywithin the overlappingportions,make it difficult to
splice together a reference section on a normalized distance
Normal Polarity
Magnetic
base.
Interval,* km
Anomaly
A procedurethat proved more tractable and testablewas to
determine a reference sectionbased on the seafloor spreading
0.00
12.14
1
pattern on a single ridge system. The South Atlantic was an
obvious

choice

as the basis of a framework

for

a reference

sectionfrom anomaly 34 to present. Seafloor spreadingin the

South Atlantic is preservedon both limbs of the ridge system,
which allows us to compensatefor ridge jumps and asymmetric
spreading,and is documentedby a large data baseof ship tracks,
aeromagneticlines, and satellite altimetry observation[Cande
et al., 1988]. In comparison,otherridge systemsare lessuseful
becauseof an inadequatelength of record (e.g., Reykjanes,and
those associatedwith the smaller plates of the Pacific), major
plate reorganization(Indian Ocean), preservationof only one
limb (North Pacific), or a more poorly resolved (North and
Central Atlantic) or less well documented (Pacific-Antarctic)
magnetic anomaly signature. For many of the same reasons
discussed above, Heirtzler et al. [1968] had also chosen the
South Atlantic

as the best ocean to use as a reference.

anomalies

20 and 23 and between

anomalies

33 and 34.

For most time intervals, however it is necessaryto fill in the

fine detail of the reversalpatternusing magneticanomalydata
from faster spreadingridges such as those in the North Pacific
and Indian

Oceans.

We

refer

to this finer

41.75
50.70
54.10
66.44

70
76.76
80A0

96.87
103.92
116.70
120.62
124.68

126.48
130.83

We used a combination of finite rotation poles and stacked

profiles from the South Atlantic to build a framework for the
time scale. Distancesbased on finite rotation poles between
selectedanomaliesare referred to as categoryI intervals; subdivisionsof theseintervalsbasedon stacksof selectedanomaly
profiles are referred to as category II intervals. For a few
portions of the time scale we judged that the category II
intervals were adequateto define the anomaly spacings,e.g.,
between

15.37
27.80
35.04

subdivision

of the

anomaly spacingsas category III intervals. A consequenceof

this procedure is that the very short wavelength anomalies,
commonlyreferredto as tiny wiggles,have not been uniformly
resolvedover the entire anomalysequence. To compensatefor
this nonuniform resolution, and in recognition of the
ambiguousorigin of the tiny wiggles, we have excluded the
shortest events (i.e., with apparent durations of less than 30
kyr) from our final compilationof anomalyspacingsand in the
resultinggeomagneticpolarity time scale.

142.49

152.32
163.49
171.00

174A7
178.38
2O3.44
213.04

223.52
229.23

240.65
242.9O
247.92
255.19

264.53
275.66

290,17
295.63
318.39
325.65
330.95
347.64

370,87
388.64

16.39
31.51
35.57
49.44
52.31
58.03
68.23
73.56
78.26

lr. 1
2
2r. 1
2An. 1
2An.2
2An.3
3n.1
3n.2
3n.3

84.68
101.42
109.60
119.74
121.30
125.35
129.08
139.37
143.15
159.16
165.16
172.34
177A9
201.13
204.51
214.28
226.81
234.25
241.35
243.94
251.38

3n.3
3An. 1
3An.2
B
3Br. 1
3Br.2
4n. 1
4n.2
4r. 1
4A
4Ar.1
4Ar.2
5n.1
5n.2
5 r. 1
5r.2
5An.1
5An.2
5Ar.1
5Ar.2
5AA

260.03
273.28
285.8O
292.24
298.45
324.87
329.38
334.88
355.45
382.45
413.88

5AB
5AC
5AD
5Bn.1
5Bn.2
5Cn. 1
5Cn.2
5Cn.3
5D

,920

CANDE AND KENT: NEW GEOMAGNETICPOLARITYTIME SCALE

and 2.5 km after removalof the long wavelengthcomponent,

also using the methodsof Schoutenand McCamy [1972]. The

TABLE 2. (continued)

causative reversal boundaries were then determined

Normal Polarity
Interval,* km

Magnetic
Anomaly

422.93

427.81

6An. 1

434.18
452.46

441.85
454.63

6An.2
6AA

461.59
468.97
472.08
477.29
490.61
498.54
506.47

463.92
469.79
475.99
483.70
495.05
501.55
509.41

6AAr. 1
6AAr.2
6Bn.1
6Bn.2
6Cn. 1
6Cn.2
6Cn.3

524.64

525.92

7n. 1

529

536.04

7n.2

543.97

547.82

552.30

555.55

8n. 1

7A

556.60
583.30

571.04
607.96

8n.2
9

616.!2
623.90
.645.65

622.16
68.29
652.56

10n.1
10n.2
11n.1

655.31

664.15

11n.2

674.26.

686.50
755.44

12
13

784.40

791.78

802.15

806.87

16n.1

15

810.93

827.67

16n.2

834.68

856.19

17n.1

859.46
867.33

865.54
872.10

17n.2
17n.3

907.31

18n.1

879.83

909.21
947.96
977.65

921.21
954.12
1006.06

1060.24
1117.55

1094.71
1130.78

1150.83
1155.75
1178.96
1185.61
1188.05

1153.90
1168.20
1184.03
1186.34
1195.35

1234.51
1257.8!
!303.81

1241.50
1262.74
1308.70

25
25
27

1325.71
1347.03
1371.84
1409.56

1341.99
1358.66
1407.22
1429.14

28
29
30
31

18n.2
19
20

21
22
23n. 1
23n.2
24n. 1
24n.2
24n.3

!481.12

1487.68

32n.1

1493.94
1539.94
1549.41

1531.81
1542.32
1723.76

32n.2
32r.1
33

1862.32

34

on the basis

of the local zero crossings.Occasionallythis proceduredid not

producea zero crossingfor a desiredanomalypick andwe would
estimatethe reversalboundaryby eye. This procedureis illustratedin Figure3.
The profiles that were averagedto obtain the categoryII
widthsare shownin Figures4a to 4i after beingdeskewedand
stretchedto a commondistanceinterval. At the top of eachset
of profiles in Figure 4 we show the average width of each
subintervalas indicatedby the vertical bars. It shouldbe noted
that the profilesin Figure4 have not beendownwardcontinued.
The downward continued versions of the profiles, used to
actuallydeterminethe verticalbars shownin Figure4, are not
shown. A profile basedon a stack of the realignedindividual
profiles is also shown, althoughpurely as a reference.The
distances between the subintervals on the stacked profile
representsour best estimateof the relative polarity widths for
the interval basedon South Atlantic data. These categoryII
distancesare given in Table 2.
FINE DETAIL FROM FAST SPREADING RIDGES

For mostof the categoryII intervalsdefinedfrom the stacked

South Atlantic profiles, it is necessaryto analyzeprofiles from
faster spreadingrate ridges in order to resolve finer detail.
Accordingly,we have compiledadditionalprofiles,mostlyfrom
the Pacific and Indian oceans, in order to define the finer scale

structureof the reversal sequence. For two portions of the

reversalsequence,betweenthe ridge axis and anomaly3A and
betweenanomalies4A and 6, we usedthe resultsof previously
published studies [Klitgord et al., 1975; Blakely, 1974,
respectively].
Our method

of construction

of a reference

section for the

relativewidthsof magneticanomaliesis thusa tieredprocessin

which successivelygreaterdetail is addedto a basic framework
basedon the finite polesfor the SouthAtlantic. This processis
illustratedin Figure 5, in which the distancesto reversalboundaries are shown along the synthetic flow line in the South
Atlantic at different stages. The left-hand column shows the
category I distances as determined from the finite rotation
poles. The next column shows the category II distancesas
determinedfrom averaginganomaly widths on profiles in the
South Atlantic. The following column showsthe category1II
distances based on the f'me detail

where it has been determined

from fasterspreadingridges. The right-handcolumnshowsthe

anomaly spacings after the removal of the nonuniformly
mappedtiny wiggles.
Tiny Wiggles

In many detailed studiesof magnetic anomaliesfrom fast

spreadingridges, linear, small scale magnetic anomalies(tiny
CategOry
I distances
arebid,Category
II distances
are
wiggles)that are clearly relatedto palcomagneticfield behavior
underlined, and Category II! distancesare plain text.
are observed. However, there are two problemsin interpreting
'1.29 km subtracted for Central Anomaly offset.
these anomalies and including them in the time scale: (1)
ambiguity in the type of dipole geomagneticfield behavior
representedby the tiny wiggles,and (2) unevenness
in temporal
The relative widths of anomalieswithin each categoryI
and spatial coverage.
interval were constrained by averaging the widths of
First, it is not clear that all tiny wiggles are due to short
subintervals on five to nine profiles that appeared to be
representativeof uniform seafloorspreading. The locationsof polarity intervals. The larger amplitude features are more
the profilesthat were usedare shownin Figure2. Eachselected obviously due to full reversalsof the field, and in some cases
profile was projected perpendicularto the local strike of the they correlate with magnetostratigraphicallyidentified short
magnetic linearions and deskewed using the phase shifting polarity intervals. However, most of the smaller amplitude
methodsdescribedby Schoutenand McCamy [1972] Profiles anomalies have not been confirmed in magnetostratigraphic
with obvious ridge jumps or other irregularities in spreading sections,and can be modeledas either due to very shortpolarity
were avoided. To pick more precisely the location of the intervals [e.g., Blakely, 1974] or else as due to longerperiod
reversal boundaries corresponding to the anomalies, the (50 - 200 kyr) intensity variations [Cande and LaBrecque,
profiles were bandpassed
and downwardcontinuedbetween1.5 1974].

CANDEANDKENT:NEWGEOMAGNETIC
POLARITY
'I'llVIE
SCALE
40

20

13,921

CHARCOT
F.Z.
27.

2'' ' ' d

'

SION

55e
5c6

F.Z.
..

r ;: 27

..

. 2524
. . 2221 20
. , ,

. ...

. ..

55

5c

20

0 Z

' f r

10

.- .. . 7

. .

24 21

18

4 r

r , , ,.

.,34

'.
'

t :

'

/
/

"

4o

. [5

'

'

'

. .

....

....

'

;, a ,'

I-

''

....

"

I
'

/ [ . '.

,,

: : 'A ' f /: '

/....//

,,

. ,

,,
I

50

D9

' '

f
40

..

-' E5 D7C6 c
z o '.

'

22'

'

. : .

6
:.'

.B2

3'0'2'4
0 /3'
,
i , '2,
,,' '

H7 ,
.e5*'

" .....

'-.

e..

',e
[

',

'

13

/'

,,.,.,.,
'20

.
10

,.. ,.,.
0

10

FiS.2. cadonofprofcs
in Sth Ariantic,
sho inFis.rcs
ro.sh4,at were
averased
toobinthecatcso

Ddistances
twccn
omacs.
Hea near
30is scfic flow]incconstrued
by setof
Second,dueto boththe unevenness
of trackcoverage
in

different areasof the ocean and the fact that the oceaniccrust has

rotation
]cs,

Cenozoic),while an equivalenthigh resolution

recordingof
Earth'smagneticfield is at leastpresentlynot availableor

recordedsometime intervalswith higherresolutionthanothers, specifically studied for other intervals. For these reasonswe
the recordof tiny wigglesis not uniform. Someintervalsare were not able to model tiny wigglesfor anomalies15 to 23 and

well covered,
eitherbecause
of exceptionally
detailedsurveys

27 to 34.

(e.g.,theNortheast
PacificNOAA surveys
of thelate 1960sand
early1970s)or because
of goodcoverage
in periods
of veryfast
spreading(e.g., the Central Indian Ocean in the early

Tiny wigglesthat werewell documented

in the analyzed
profiles were modeled as short polarity intervals as a
convenient
meansof identifyingtheirpositionwithin the well

27

2829

30

31

32

REVERSAL BOUNDARIES

Fig. 3. Exampleof treatment

of profiles. The originalprofileis
projectedperpendicular
to the strikeof the lineationsCoottom),
bandpassed
anddownward
continued
between
1.5and2.5km(second
to
the bouom),andthendeskewed
anddetrended
(secondfromthe top).
Thereversal
boundaries
(top)aredetermined
fromthezerocrossings
of
the deskewedprofile.

Fig.4. Average
distances
between
anomalies
in theSouthAtlantic,
corresponding
to thecategory
II distances,
areindicated
by thevertical
barsalongthe horizontalline throughthe stackedprofile. These

DESKEWED

distances
weredetermined
by averaging
the widthsof the subintervals
DOWNWARD

CONTINUED

ontheindividual
profiles.Eachprofilewastreated
asin theexample
in
Figure3 in orderto determine
thezerocrossings
shown
by thevertical
bars. Hand-entered
points,whenthe treatmentfailedto obtaina zero

ORIGINAL

crossing
(veryinfrequent),
areindicated
byvertical
barsthatarehalfthe
heightof thenormalverticalbars. The profilesas shownherehave
onlybeendeskewed,
not downward
continued,
although
downward
continued
profiles
wereusedto determine
thevertical
bars.Figures
4a

PROFILE

through4i correspond
to theninecategory
I intervals.
AXIS

2A

3A

4A

5A

5AC

5B

25
km
i I600nT

5C

5D

5E

I/",,

I/q,

I/

6A

I,-, /ql

'wV

6B

2_5
km

6C

d
25km
1300nT

I..

I STACK

F'-Jl

c6

lO

11

3oo
nT

12

S^CK
25km
1300nT

CANDE AND KENT: NEW GEOMAGNETIC POLARtrY TIME SCALE

13,923

' 1300
nT
25km

24

25

26

27

28

29

--'--'--
t300
nT
25km

Fig. 4. (continued)

13,924

CANDE AND KENT: NEW GEOMAGNE-'EIC

POLARITYTIME SCALE
CATEGORY
I

CATEGORY

TI

Ti

'FFI

TABLE 3. Intervalsof Uniform CrustalMagnetization

Corresponding
to Tiny WigglesFrom Anomalies1 to 13

IOO0

and From

Anomalies

24 to 28

Tiny
Wiggle

Interval,* km
i

2CXD

1200

400

1400

(Km)

(Km)

600

1600

800

1800

IOO0

Fig. 5. Distancesto anomaliesalong the syntheticflow line in the

SouthAtlantic. The distancesare built up from a combinationof finite
rotationpoles (categoryI), averagedwidths of anomaliesin the South
Atlantic (category II), and detailed studies of fast spreadingrate
anomalies(categoryIll). The right-handcolumn showsthe anomaly
spacingsafter the removalof the nonuniformlymappedtiny wiggles.

7.63
18.82
38.75
150.97
184.42
189.93
195.80
360.31

7.81
18.99
39.10
151.30
184.60
190.46
196.15
360.94

ln-1
lr.2r- in
2r.2r-1
4r.2r-1
5n.2n-1
5n.2n-2
5n.2n-3
5Dr-1
6Cr- 1

518.28

518.56

539.96

540.37

565.72
592.71
598.64
611.80
635.65

566.02
593.18
599.11
612.11
636.00

8n.2n-1
9n-1
9n-2
9r-1
10r- 1

639.95
668.93
694.06
700.67

640.13
669.30
694.56
700.92

10r-2
1 lr- 1
12r-1
12r-2

710.53

711.03

12r-3

715.15
719.64
726.50
730.61
735.10
748.12
758.91
764.15
771.73
778.14
1198.63
1200.82
1204.11
1206.30
1209.31
1213.97
1217.53
1220,54
1222.19
1225.47
1229.58
1245.52
1247.76
1249.77
1253.12
1255.13
1269.63
1277.28
1280.08
1282.64
1287.99
1292.58
1296.16
1344.19

o
o
-

715.39
719.89
726.99
730.86
735.35
748.56
759.34
764.36
772.16
778.57
1198.77
1200.96
1204.25
1206.44
1209.45
1214.11
1217.67
1220.68
1222.32
1225.61
1229.72
1245.75
1247.98
1249.88
1253.23
1255.35
1269.88
1277.53
1280.21
1282.76
1288.25
1292.71
1296.28
1344.58

7r-1

12r-4
12r-5
12r-6
12r-7
12r-8
13n-1
13r- 1
13r-2
13r-3
13r-4
24r-1
24r-2
24r-3
24r-4
24r-5
24r-6
24r-7
24r-8
24r-9
24r-10
24r-ll
25r-1
25r-2
25r-3
25r-4
25r-5
26r-1
26r-2
26r-3
26r-4
26r-5
26r-6
26r-7
28r-1

establishedreversal sequenceand becausethis producedmore

realistic appearing models of the observed anomalies.
However, in converting the distance reference section into a
time scale,we tabulatedthesepresumedshortpolarityintervals
separatelyto emphasizeboth the uncertaintyin interpretation
(dipole field intensityvariation vs. polarity reversal)as well as
the temporal nonuniformity of the existing record. Tiny
wiggles that convertedto polarity intervalsshorterthan 30 kyr
are listed separatelyin Table 3. We designatethe tiny wiggles
in Table 3 using a system that parallels the nomenclature
adopted for the larger anomalies (Table 2), except that the
apparentpolarity of the interval must be inferred to be opposite
that of the anomalyit occursin. This nomenclatureis discussed
in the appendix.
Many tiny wiggles, even on fast spreadingridges, are
'1.29 km subtracted for Central Anomaly offset.
modeledas bodiesonly 0.5 to 1 km wide. When theseintervals
are interpolated into the South Atlantic flow line, for the
purposes of constructing the time scale, they occasionally systemsinto the South Atlantic categoryII distanceframework.
convert to bodies as narrow as 0.1 kin. Consequently,we Figure 6 shows the location of these data in the context of the
global ridge system.
presentthe distancesin Tables 2 and 3 to a resolutionof 0.01
The central anomalyto anomaly3A. The detailedspacings
km, so that theseintervals are not inaccuratelyportrayeddue to
roundoff errors.
from Klitgord et al.'s [1975] studyof deeptow andsurfacedata
from various ridges in the Pacific were used to define the
spacings from the axis to anomaly 3A. The spacingsof
Fine Details of the ReversalSequence
reversalswithin this interval were interpolatedinto the South
Below are descriptionsof the magneticanomalydata usedto
Atlantic spacings(Figure 7, bottom) as is shown in Figure 7
insert detail of the reversalsequencefrom faster spreadingridge (middle). Four tiny wiggles have been identified from marine

CANDE AND KENT: NEW GEOMAGNEC POLARITY TIME SCALE

72N

120E

150E

180

150W

120W

90W

60W

30W

13,925

30E

60E

90E

120E
72N

60N

60N

FIG.

14

SON

FIG. $9

50N

OS

50S

60S

60S

72S

72S

120E

150E

180

150W

120W

90W

60W

:50W

:50E

60E

90E

120E

Fig. 6. Referencemap showinglocationof datamapsusedin thisstudy.

2A

WILSON & HEY (1981)

REA & BLAKELY (1975)

KLITGORD

SOUTH

ET AL. (1975)

ATLANTIC

2('3 km

Fig. 7. Distances
between
theaxisandanomaly
3A, builtup fromthecategory
II distances
in theSouthAtlantic(bottom),
thestudyof Klitgordet al. [ 1975](middle),andothershortevents(top).

averagedthe widthsof anomalieson four aeromagnetic

profiles
from the Chile Ridge (Figure 8) [Tebbenset al., 1990]. We
chosetheseprofilesbecausethe areal coverageis good and the
spreadingpattern is very uniform (Figure 9). The stacked
profile (Figure 9, top) reveals several small scale anomalies
betweenanomalies3A and 4A that may be indicativeof short
polarity intervals, including one anomaly on the old side of
anomaly4 that has been previouslyrecognized(4.3), but also
small anomalieson the youngshoulderof anomaly4A, on the
young shoulder of anomaly 4, and on the old shoulder of
Blakely [1975] identifieda tiny wigglebetweenthe Jaramillo anomaly 3B.
and anomaly 2, which correlates with the Cobb Mountain
We stacked two profiles from the east flank of the fast
subchron [Mankinen et al., 1978; Mankinen and Gromme,
spreading
EastPacificRise(Figure10) to furtherinvestigate
the
1982; Clement and Kent, 1987]. The compositereversal small-scale anomalies seen in the Chile Rise stack. These
sequence
includingthesefour tiny wigglesis shownin Figure7
profiles and their stack (Figure 11) reveal that the shoulder
(top).
anomaliesare indeedfield relatedtiny wiggles. More speculaAnomaly3A to anomaly4A. The sequence
betweenanomaly tively, we point out that thereis a suggestion
of an evensmaller
3A and 4A is not well constrainedfrom publishedstudies. tiny wiggleon the youngshoulderof anomaly3B, althoughwe
Talwani et al. [ 1971] calculateda setof relativeanomalywidths have not included it in our final model.
based on data from the ReykjanesRidge, but due to the slow
We havemodeledthe four tiny wigglesseenin the Chile Rise
spreadingrate the details of the polarity sequencewere not stackand confirmedin the EPR stackas due to shortpolarity
clearly resolved. To determine the relative spacings, we intervals. The spacingsfor the South Atlantic framework,the
magnetic studies in this interval. Wilson and Hey [1982]
recognizeda distinctivetiny wigglewithin the Centralanomaly
which correlates with the Emperor subchron [Ryan, 1972;
Championet al., 1981] (but see Championet al. [1988]). Rea
and Blakely [1975] identified two tiny wiggles between
anomalies2 and 2A: the younger feature correlateswith the
Reunion subchron(s)[Gromme and Hay, 1971] and the older
correlates with the "X"-anomaly [Heirtzler et al., 1968], for
which there is only weak magnetostratigraphicevidence
[Mankinen and Dalrymple, 1979]. More tentatively,Rea and

13,926

CANDE AND KENT: NEW GEOMAGNETICPOLARITYTIME SCALE

-90

-lOO
-38

-80

' II !i--"-"""
3
PMgOSW

3I

'

-70

'

'

-38

PMSOSE

-46

-46

-90

-80

-70

Fig. 8. Locationof aeromagnetic

profilesfromtheChileRidgeusedto constrain
therelativewidthsof anomalies
3A to 4A in
Figure 9.
-120

3A

3B

:.::.

2_Skin
I500nT
N UN

-100_10

4A

t/

PMgOTW

, 'VLCN8

I/

/q

7,

PM90BW

-3o

-30

-100

-' 20

10. Locationof two profilesfrom the East PacificRise shownin

PM90BEFig.
Figure 11.

Fig. 9. Averagedwidthsof subintervals

betweenanomalies3A and4A,
as indicatedby vertical bars along top horizontallines. Four profiles
from the Chile Ridge, showndeskewed,were averagedto obtainthe
widths.

Chile Rise constraints,and the full reversal sequenceincluding

the tiny wiggles, are shownin Figure 12 (bottom,middle and
top, respectively).
Anomaly 4A to anomaly 6. For this time interval, we have

taken the spacingsdeterminedby Blakely [1974], who stacked

14 profiles from the North PacificNOAA survey,and inserted
them into our South Atlantic framework. In Figure 13 we show
the South Atlantic spacings (bottom) and the composite

I,-,I I.

L,,I

,,

,,"'q VLCN8

IO4
' ',,,,/1 ^ 1/ P7303

reversalsequence
basedon the Blakely [1974] pattern(top).
Fig. 11. Deskewedprofiles from the East Pacific Rise showing
Anomaly 6 to anomaly8. For this interval we stackedtwo correlatabletiny wigglesnear anomalies3B and4.
profilesfrom the NorthPacificNOAA surveylocatedjust to the
west of the profilesanalyzedby Blakely [1974]. The profiles
were selected(see Figure 14 for location)to avoid the complex 10 and 11. We have also identified eight tiny wiggles in the
pattern of propagatingrifts apparent in this region [e.g., stack and have modeled them as discrete short polarity
Atwater and Severinghaus,1989]. The profilesand their stack intervals. The South Atlantic framework, the detail from the
stack,and the shortpolarity intervalsas we have modeledthem,
are shownin Figure 15.
The stack delineatesthe relative width of the many short are shownin Figure 18 (bottom,middle and top, respectively).
Anomaly 12 to anomaly 15. From the same generalarea of
anomalieswithin this interval. These apparentreversalshave
been insertedin the SouthAtlantic framework(Figure 16, center the NOAA survey(Figure 14), we haveselectedandstackedfive
and bottom). In addition, we note a small anomaly on the profiles crossinganomalies12 to 15 (Figure 19). The stack
on both of the
youngshoulderof anomaly6B, a tiny wigglein the centerof revealsa patternof tiny wiggles superimposed
anomaly6B, and a tiny wiggle betweenanomalies6C and 7. long negative anomalies. The eight tiny wiggles between
We have modeled these anomalies and included them in the

reversalpattern(Figure 16, top).

Anomaly 8 to anomaly12. We have stacked7 profilesfrom
theNOAA surveybetweenanomalies
8 and12 (seeFigure14 for
location). The stack of theseprofiles (Figure 17) providesthe
spacingfor the intervalsbetweenthe twin peaksof anomalies

anomalies12 and 13 were describedand modeledby Cande and

LaBrecque [1974]; we recognizethe sametiny wigglesin our
modelpresentedhere. Betweenanomalies13 and 15 we have
identifiedand modeledfour tiny wiggles,and we alsoidentify a
tiny wiggle within anomaly 13. (It is interestingto note that
Heirtzler et al. [1968] included four short polarity intervals

3A

13,927

3B

+ EAST

CHILE

SOUTH

PACIFIC

RISE

RIDGE

ATLANTIC

10 km

Fig. 12. Distancesbetweenanomalies3A and 4A basedon a combinationof the categoryII South Atlantic distances
(bottom),the averagedChileRidgeprofiles(middle),andmodelsof the tiny wigglesobserved
in the EastPacificRiseprofiles
(top).

4A

5A

5AC

5B

5C

5D

5E

U+BLAKELY
(1974)

SOUTH
ATLANTIC
I

30

km

Fig. 13. Detailedspacingsbetweenanomalies4A and 6 basedon the SouthAtlanticcategoryI intervals(bottom),and the

detailedstudyof the NortheastPacificNOAA linesby Blakely [1974] (top).
-170
50

-160

-150

- 140

-130
50

la

IDOE10

4O

4O

30

-170

-160

/ l \

- 150

-140

3O

-130

Fig. 14. Locationof profilesin the NortheastPacific usedto constrainthe fine details of anomalies6 through20 and
anomalies

28 to 32.

5E

6A

6B

6C

V W v"

7A

...nU u
I 50kmI

/l/v

UV V Vur'

. n IIL4i/
hH

500nT

U V

N I IDOE!

Fig. 15. Two NOAA lines from the NortheastPacific that were averagedto constrainthe widths of subintervalsbetween
anomalies 5E and 8.

13,928

CANDEANDKENT:NEWGEOMAGNETIC
POLARITYTIME SCALE

5E

6A

II

6B

6C

7A

+TINY
WIGGLES

l-l__i-I I_I-L]J

+NORTH
PACIFIC
SOUTH

ATLANTIC

.30

km

Fig. 16. Distancesbetweenanomalies5E to 8 as constrained

by a combinationof categoryH distancesfrom the South
AtlanticCoottom),
the averageof the two linesshownin Figure15 (middle),anddetailedmodelingof the tiny wiggleswithin
anomaly6B, betweenanomalies
6C and7, andbetweenanomalies
7 and7A (top).
8

10

11

12
between

STACK

anomalies

13 and

15 in their

time

scale

and the

youngestof them was called anomaly 14.) We show the South

Atlantic framework and the reversal sequencebased on it in
Figure 18 (bottom and middle) and the short polarity intervals
basedon the North Pacific data in Figure 18 (top).
Anomaly 15 to anomaly 20. The fine detail within anomalies
15 to 20 was constrainedfrom five profiles from the North
Pacific locatedto the west of the NOAA surveyarea (Figure 14).
We showthe profilesandtheir stackin Figure20. The stackwas
used to constrainthe relative widths of the anomalies(Figure
21, middle) and as a guide to model the fine detail within
anomaly17 (Figure 21, top). There are almostcertainlyadditional cotrelatable tiny wiggles within this sequence. For
example, within the young part of anomaly 18 there is a tiny
wiggle observed on the two lowermost profiles in the stack.
However,we have not modeledthe tiny wigglesin this sequence
as we have for anomalies8 through15 becausethe profiles are
more widely separatedthan in the NOAA survey area and the
stack is not as definitive a record of the reversal history. For
example, the tiny wiggles within the negative intervals on
either side of anomaly 19 are almost as distinct as the tiny
wiggle within anomaly 18, yet we would feel uncomfortablein
modeling it as a true field related event based only on these
profiles. We prefer to leave the tiny wiggle record blank and
note that in any casethe featuresin questionwould be too short

50
km
I5OOn'l'

// /f-/ IDOE3

, ,,

io.4
IDOE5

IDOE6

IDOE?
IDOE8

IDOE9

Fig. 17. Seven NOAA lines from the Northeast Pacific that were
averaged to constrain the widths of subintervalsbetween anomalies 8

to be included in the final time scale.

Anomaly 20 to anomaly 23. The reversalsequencein this

time interval consists of several long reversed and normal
polarity intervals. The SouthAtlantic spacingsare adequateto

and 12.

R II

II

10

11

12

15

15

lll-IJIII- II
+ NORTH PACIFIC

SOUTH
i

ATLANTIC

.30 km

Fig. 18. Distancesbetweenanomalies8 and 15 as constrained

by a combination
of the categoryH distances
form the South
Atlantic Coottom),
the averageof the NOAA lines shownin Figures16 and 18 (center),and detailedmodelingof the tiny
wigglesobservedin the stacks(top).

CANDE AND KENT: NEW GEOMAGNE'rlc POLARITY TIME SCALE

model this interval and no exceptional surveys allow a

systematic
searchfor tiny wiggleson fasterspreading
ridges.
Anomaly23 to anomaly28. We have stackedtwo profiles
from the southflank of the CentralIndianRidge(Figure23, see
12

t ttttt

15

13,929

Figure22 for location)that recordthe reversalsequencein this

time interval at the fast spreadingrate of roughly 60 mm/yr
(half rate). We have used the stack to constrain the relative

widthsof the polarity intervals(Figure24, middle).

There are numeroustiny wiggles in this time interval; a
relatively long one (-50 k.y.) has been previouslymodeled
within anomaly24 [Schlich, 1975]. In addition,from a detailed
studyof a largenumberof profilesfrom the Indian Ocean[Cande

STACK

and Kent, 1992],we recognizean essentially

continuous
pattern

of tiny wiggles between anomalies24 and 27. We find that

,50
km
1,500
nT

almost

IDOE
10

all of the small

scale features

observed

in the two

profilesin Figure23 are observedon otherprofilesin the Indian

Oceanandarerecognizedin a stackof theprofiles. In Candeand
Kent [1992] we modeledthe patternof very short polarity
intervalsthat simulatesthe small scaleanomalies;this pattern
is shownin Figure 24 (top).
The recognitionof the continouspatternof tiny wiggles
betweenanomalies24 and 27 has led us [Candeand Kent, 1992]
to developan alternativemodelfor their interpretation.Neither
of the existing models for tiny wiggles is particularly
appealing. If all tiny wigglesare due to shortpolarityevents
[e.g.,Blakely, 1974], thenit wouldrequirea sixfoldincreasein
reversal frequencyto accountfor the large number of tiny
wigglesbetweenanomalies24 and27. Alternatively,the Cande
and Labrecque[1974] modelof longperiodintensityvariations
requiresfluctuationsof the dipole field at periods(50 to 200
kyr) substantially longer than generally attributed to core

IDOE12

._._._

IDOE
13
IDOE14

Fig. 19. FiveNOAA linesfromtheNortheast

Pacificthatwereaveraged processes(~10 kyr). In Cande and Kent [1992], we show that
to constrainthe width of anomaly13 and to determinethe locationof
the cotrelatabletiny wigglesbetweenanomalies13 and 15.

15

15

16

tiny wigglescan be attributedto random,shortperiod(2 to 20

kyr) intensityvariationsof the dipolefield which appearas 15

17

18

19

F-^r}

STACK

5
aT
50 km

Yq?0
1

P689-A
P689-

P?00
Pi. 20. Pie profilesfromtheNoN Pacificat wereaeed
d 20, assowny e ecal ars onthetopodn]

15

16

17

18

19
+

NORTH

SOUTH

PACIFIC

ATLANTIC

50 km

Fig.21. Distances
between
anomalies
15and20 asconstrained
bya combination
of thecategory
II distances
fromtheSouth
Atlantic
(bottom),
andthedetailfromtheaveraged
widths
of thesubintervals
ontheNorthPacific
profiles
(top).

13,930

CANDE AND KENT: NEW GEOMAGNEC POLARITY TIME SCALE

to 25 km wavelength anomalies at the sea surface due to the

bandpassingeffect of the "Earth filter" [Schoutenand McCamy,
1972]. This suggeststhat tiny wiggles do not require an unusal
type of field behavior and, therefore, that they are not
unexpectedas a backgroundsignal in fast spreadingrate, high
resolution,marine magnetic anomalyrecords.
Anomaly 28 to anomaly33. We have constrainedthe relative
widths of the polarity intervals in this time period using
profiles from the North Pacific. The relative widths of
anomalies in the North Pacific varies considerably between
spreading corridors. Consequently, instead of taking the
average of a stack we selected short sections of individual
profiles: Geosecs-A(GECSA) from North of theMurray Fracture
Zone for the relative widths of anomalies 28 and 29, and

Pioneer 7103 (P7103-1) from just south of the Surveyor

FractureZone for anomalies30 to 32. Theseprofilesare shown
in Figure 25; see Figure 14 for location. Except for a widely
recognizedshort normal polarity interval between anomalies32
and 33 and a tiny wiggle identified by Schlich [1975] between
anomalies28 and 29, we have not modeledany tiny wigglesin
SO

60

70

-EO

-EO

this time interval. The resulting reversal pattern is shown in

Figure 26.
Anomaly33 to anomaly34. We have not modeledany fine
detail in this time interval, as these distinctive, long
wavelength anomalies are well constrainedfrom the South
Atlantic

alone.

Error Analysis

Errorsin the widthsof the categoryI and categorylI intervals

can be estimatedin a straightforwardfashion. We estimatethat
the confidence ovals and the errors in the rotation angles
(estimatedby Cande et al. [1988] to be roughly 0.05 to 0.1)
for the finite rotation poles used to constrain the category I
intervalstranslateinto an error of roughly 10 km in the distance
along the flowline, or an accuracyof 1 to 3% dependingon the
width

of the interval.

Errors in the widths of the categoryII intervalsare calculated

from the statistics of the individual

estimates that were used to

determinethe averagewidths. By assumingthat the individual

estimatesare normally distributedwith a standarddeviation o,
aboutthe meanwidth, , of N observations,
the true mean, IXis
expectedto lie with 95% confidencewithin the limits

_ 1.96o<!x< + 1.96o

-3O

In Table 4 we give the mean width, the number of

observations,the 95% confidenceinterval and the percenterror
that the 95% confidence interval represents,for all of the
categoryII intervals. The percent error ranges from 1.1% for

-3O

the interval between anomalies 13 and 15, to 17% for the

interval between anomalies 23 and 24. The average error is

roughly 7%.
Becausethe categoryIII intervalsare basedon either one or
two carefully selectedprofiles or else stacksof profiles from a
small area, it is not possibleto make a significantestimateof
the error in these intervals.
-4O

-4O

Adjustmentto the Central Anomaly and the CompositeSequence

-50

To accountfor the finite width of the emplacementzone of

magnetizedoceanic crust, a small but systematicadjustmentof
the anomaly distancesis requiredbefore the compositereversal
sequenceis calibratedto generatea time scale. The width of the
transition zone over the Brunhes/Matuyama boundary in the
South Atlantic as inferred from the shapes of magnetic
anomalies is estimated at about 3 km [Carbotte et al., 1991].
Since the outward displacementof polarity boundariesdue to
this effect is expectedto be about half the transitionzone width
[Atwater and Mudie, 1973], it is necessaryto subtractroughly
1.5 km from all the anomaly distances in the composite

-50
50

70

Fig. 22. Location of profiles in the Indian Ocean that were used to
constrain

the widths

of anomalies

22 to 28.

25

22

25

24

26

27

I STACK
I I1500
nT
100 km

V'".,
v

GAI,03

MDF16

Fig. 23. Two deskewedprofiles from the Indian Ocean that have been averagedto obtain the widths of the sub-intervals
betweenanomalies22 and 28. The tiny wigglesbetweenanomalies24 and 27 are remarkablycorrelatable.

CANDEAND KENT:NEW GEOMAGNETIC

POLARITYTIME SCALE

22

25

24

25

26

13,931

27

flllll Illllllllll I IllIll

+ TINY

WIGGLES

+ INDIAN

SOUTH

OCEAN

ALANTIC

50 km

Fig. 24. The distancesbetweenanomalies22 and 28 as constrainedby a combinationof the categoryII distancesin the South
Atlantic(bottom),the averagedwidthsof the subintervals
on the IndianOceanprofiles(middle),anddetailedmodelingof the
tiny wigglesbetweenanomalies24 and 27 (top).

27 28 29

51

52

55
GECSA

500
nT

P7103-1
100km

Fig. 25. Two profilesfrom the North Pacificusedto constrainthe widthsof the subintervals
betweenanomalies28 and 33.

28

29

30

31

32

WIGGLES
I-I/

+ NORTH

SOUTH
I

30 km

PACIFIC

ATLANTIC

Fig. 26. Distancesbetweenanomalies28 and 33 as constrainedby a combinationof the SouthAtlantic categoryII distances
(bottom),the North Pacificprofile (center),and modelingof the shorteventbetweenanomalies32 and 33 and Schlich's
[1975] tiny wiggle betweenanomalies28 and 29 (top).

sequence. Failure to do so leads to an unreasonablyold age of

the Brunhes/Matuyama boundary of well over 0.8 Ma. The
actual amount of the correction that we used, 1.29 km, was

calibrationof the geomagneticpolarity intervalsto make a time

scale

is thus

reduced

to the

determination

of

the

seafloor

spreadinghistory of the South Atlantic.

Previousgeomagneticreversal time scale calibrationmethods
determined by forcing the age of the Brunhes/Matuyama
boundaryto equal a predeterminedvalue as discussedlater. The assumedone or more extendedintervals of constantspreading
consequenceof this correction for Central Anomaly offset is rate in the South Atlantic, ranging from over practically the
that the half-width of the Central Anomaly is reducedfrom about entire ridge crest anomaly sequence(HDHPL68; LKC77) to the
13.4 km to 12.2 km in the South Ariantic reference flow line.
nominal duration (-6 m.y.) of geologic stages [Lowrie and
Alvarez, 1981] (LA81). The assumptionof constantspreading
AGE CALIBRATIONOFCOMPOSITE
SEQUENCE
rate over extended time intervals is clearly an oversimplification and is no longer necessarywith the availability of more
We have derived a compositegeomagneticpolarity sequence numerousage calibration data. On the other hand, strict linear
along a synthetic flow line for the South Atlantic ridge system interpolation can result in frequent and abrupt changes in
(right-hand column of Figure 5; Tables 2 and 3). Age apparentspreadingrate that are simply artifactsof inaccuracies

13,932

CANDEAND KEwr: NEW GEOMAGNETICPOLARITYTIME SCALE

in closely spacedcalibrationtie points, as was pointedout for

example,by Harland et al. [ 1982].
In our initial efforts to identify long profiles from different
ridge systemsthat could be used as referencesequences,ridgeridge comparisonsinvolving the South Atlantic showed the
fewest kinks or sudden shifts in relative spreading. This
suggestedthat South Atlantic spreadingwas relatively uniform

and continuous. Accordingly, we assumefor the purposesof

temporal calibration a smoothly varying (but not necessarily
constant)rate of spreadingfrom anomaly 34 to the ridge axis
for the South Atlantic ridge system. In practice, we chosenine
more or less evenly distributed age calibration points, and fit
them with a cubic spline approximationto interpolatethe age
of polarity intervals.

TABLE 4. Category II Statistics

Mean

Anomaly

Width, km

95% Confidence

Interval, km

% Error

In

13.425'

0.784

6.3

lr

15.662

1.782

11.3
4.1

13.953

0.576

2A-3n. 1

28.809

2.118

7.3

3r2-3r

26.314

13.3
6.0

29.605

3.524
1.789

25.838

3.226

12.4

20.020

2.142

10.7

51.186

2.355

4.6

5A-5AB

41.008

2.419

5.9

5AC-5AD

25.639

2.282

8.9

3A-3B
4

4A
5

5B

28.217

1.298

4.6

5C

29.255

4.154

14.2

5D

23.224

2.392

10.3

5E

17.775

1.653

9.3

34.292

3.223

9.4

6A

18.909

1.456

7.7

6AA

27.126

2.740

10.1
12.7

6B

21.643

2.749

6Cn

18.794

1,428

7.6

6Cr

17.881

1.502

8.4

12.1

7n
7r-7A

8n

8.732

1.057

16.297

0.652

4.0

18.714

1.273

6.8

8r

12.194

0.988

8.1

9n

24.739

3.018

12.2

9r

8.176

0.589

7.2

10n

12.146

1.494

12.3

13.2

10r

17.268

2.279

11n
11r

18.588
10.023

9
9

1.747

9.4

1.383

13.8

12n

12.261

1.238

10.1

12r

56.206

3.091

5.5

13

41.766

O.459

1.1

15

17.754

0.728

4.1

16

32.53O

1.984

6.1

17

45.142

2.754

6.1

18n

41.386

2.856

6.9

18r

26.744

0.749

2.8

19

29.692

2.108

7.1

20n

28.413

2.046

7.2

20r

54.180

3.576

6.6

21n

34.466

2.895

8.4

21r

22.843

2.924

12.8

22n

13.232

1.059

8.0

22r

20.046

2.385

11.9

23

28.137

4.868

17.3

24

55.547

5.221

9.4

25

23.297

2.260

9.7

26

46.001

2.8O6

6.1

27

21.900

3.022

13.8

28

46.136

4.475

9.7

30-31n
31r
32

57.297
51.976
68.291

8
8
8

5.730

10.0

3.430

6.6

5.736

8.4

33n

174.348

5.753

3.3

33r

138.564

5.820

4.2

*Not corrected for 1.29 km Central Anomaly offset.

CANDEAND KENT: NEW GEOMAGNEC POLARITYTIME SCALE

13,933

Harland et al. [1990] recently analyzedthe availablebody of

diagnosticradioisotopicage data in the form of chronograms
for geologic stage boundaries. In essence, a chronogram
portrays a measure of the overall inconsistencyin a set of
constrainingradioisotopicage data as a function of trial ages
for a particular boundary;the best chronogramestimateof the
boundary age is that which minimizes the measure of
inconsistencyamong the dates weighted according to their
individual analytical errors. In principle, chronogramestimates
thusembodythe influenceof all the availableradioisotopicdata
and consequently should be robust.
In practice, the
chronogramsoften do not provide acceptableage control due
mostly to the lack of an adequatenumber and distributionof
reliable dates. Consequently,Harland et al. [1990] did not use
chronogram estimates from virtually the Priabonian to
B artonian Stage boundary in the Eocene to the Campanianto
SantonianStage boundaryin the Late Cretaceousfor calibration
of the GTS89 geomagneticpolarity time scale.
We selected magnetobiostratigraphicallywell-correlated
calibration points spaced at approximately 10 m.y. intervals,
comparable in temporal separationto the category I intervals

basis of new, high-precisionradioisotopicdating [Swisher and

Prothero, 1990; Prothero and Swisher,1992]. In the following
section, we describedetails of the calibration points using a
chronnomenclatureas definedin the appendix.

[1990].

mination
well-defined
40Ar/39ArdatesfromlavasfromMaui

Description of Calibration Points

1. After the zero-age ridge axis, the youngestcalibration

point in virtually every geomagneticpolarity time scale since
and includingHDHPL68 hasbeen for the old end of anomaly2A
(= C2Ar(0.0)), correlative to the Gauss/Gilbert boundary.
Mankinen and Dalrymple [1979] give a chronogramestimateof
3.40 Ma for the Gauss/Gilbertboundary. Recently, however, it
has been possibleto date high resolutionclimate recordsfor the
Plio-Pleistoceneby assumingthat their variability was forced
by the well-known variationsin the Earth'sorbital parameters.
Astronomicaltime scalesdevelopedfrom climatic records from
the equatorial Pacific [Shackleton et al., 1990] and the
Mediterranean region [Hilgen, 1991a, b] reveal that the
conventional radiosotopic age estimates for geomagnetic
reversal boundariesin the Plio-Pleistoceneare consistentlytoo
used to construct the overall framework for South Atlantic seayoung by about5 to 7%. For example,Hilgen [1991b] derives
floor spreading. The selectedtiepointsin the Neogene largely an astrochronologic age of 3.58 Ma for the Gauss/Gilbert
conform to the chronogram estimates from Harland et al.
boundary. Similarly, the astrochronologic age for the
[1990], whereas for the tiepoints in the Paleogene, we were Brunhes/Matuyama boundary derived by Shackleton et al.
guidedby a recent assessment
of correlationsand age estimates [1990] is 0.78 Ma as comparedto the chronogramestimate of
by Berggrenet al. [1992]; Late Cretaceousage assignmentsare 0.73 Ma given by Mankinen and Dalrymple [1979]. In a
similar to those in Berggren et al. [1985b] and Harland et al.
remarkable confirmation of this astrochronologic deterThe nine selected calibration points (Table 5) are a
distillation of a much larger number of individual radioisotopic
datesas describedbelow. The tiepointsreflect a preferencefor
thosedata which can be tied to the magneticanomalysequence
via marine magnetobiostratigraphic correlations and
constraints from biostratigraphic correlation of sediments
overlyingoceanicbasement[Candeet al., 1989]. An exception
is the tiepoint at the Cretaceous/Paleogene
boundaryfor which
the iridium anomaly provides a means of precise correlation
between

the marine

record

and the nonmarine

sediments

from

which the radioisotopicdates have been obtained. In general,

though, data from terrestrial (i.e., nonmarine) sections are
generallynot sufficiently independentfor time scale calibration
becausethe radioisotopicage informationitself tendsto be used
as a basis for detailed correlation to the marine magnetic
anomalysequence. A casein point are the two Oligoceneash
layers in terrestrial deposits in western North America that
provided important age calibration constraints for the
Paleogene part of BKFV85, but whose correlation to the
polarity sequencehas been successivelyreinterpretedon the
TABLE 5. Age Calibrationsfor GeomagneticPolarity Time
Scale

C2An(0.0)
CSBn(0.0)
C6Cn,2r(0.0)
C13r(.14)
C21n(.33)
C24r(.66)
C29r(.3)
C33n(.15)
C34n(0.0)

distance,* km

41.75
290.17
501.55
759.49
1071.62
1221.
1364.37
1575.56
1862.32

al., 1991a,b; Walker et al., 1991' Tauxe et al, 1991] also

suggeststhat the earlier conventionalK/At dates that have been

generally used to estimate the ages of the most recent

geomagnetic reversal boundaries are the main cause of the
discrepancies, rather than inaccuracies in the fundamental
radiometricdecay constants.
Acknowledgingthe break with traditionin attemptingto take
into accountthe new information, we use a calibrationtiepoint
at the Matuyama/Gaussboundary(= C2An(0.0)) set to 2.60 Ma,
as determinedfrom astronomicalcalibrationof two independent
setsof data by Shackleton et al. [1990] and Hilgen [1991b].
This compareswith a chronogramestimateof 2.48 Ma for the
Matuyama/Gauss boundary determined by Mankinen and
Dalrymple [1979]. We alsoused the astrochronologic
estimate
of 0.78 Ma for the Brunhes/Matuyamaboundary to determine
the amount of correction for the outward displacementof the
Central Anomaly (1.29 kin) as discussedearlier.
2. Most recent time scales(e.g., BKFV85, GTS89) have a
calibration point at anomaly 5 (= chron C5n). However, in

view of the continueduncertainty(discussed

below) regarding

South Atlantic

Chron

which lie within the Brunhes/Matuyamatransitiongive an age

of 0.783 Ma [Baksi et al., 1992]. The precise agreementnow
emerging between the astrochronologicages and new high
precisionradiometric agesfor the last 5 m.y. [see also Baksi et

Age, Ma

2.6
14.8
23.8
33.7
46.8
55.0
66.0
74.5
83.0

'1.29 km subtracted to accountfor Central Anomaly offset.

what is the best age estimatefor chron C5n, we prefer to use

insteada calibrationpoint lower in the Miocene. Accordingly,
we assignan age of 14.8 Ma to the youngerend of chronC5Bn
(= C5Bn(0.0)). This is based on radioisotopicage constraints
on the correlativeN9/N10 foraminiferazone boundaryof Miller
et al. [1985] and Berggrenet al. [1985b], as estimatedin Japan
(14.6 +_0.4 Ma [Tsuchiet al., 1981]) and in Martinique(15.0 +
0.3 Ma [Andreieft et al., 1976]). This calibration agreeswell
with derived agesfor chron C5Bn(0.0) of 14.87 Ma in BKFV85
and 14.98 Ma in GTS89, as well as a chronogramestimate of
14.6 Ma [Harland et al., 1990] for the approximatelycorrelative
Serravallian/Langhian boundary.
3. The Miocene/Oligocene(Aquitanianto Chattian)boundary
can be correlated to the middle part of chron C6Cn (=
C6Cn.2r(0.0)) [Berggren et al., 1985b], and has a chronogram
estimate of 23.8 Ma (five dates listed for the Aquitanian, 20

13,934

CANDE AND KENT: NEW GEOMAGNE-'rlC

POLARITYTIME SCALE

56.5Ma. Onthebasis
of new40Ar/39Ardates
onvolcanic
ash

datesfor the Chattian) accordingto Harland et al. [1990]. This

is virtually identicalto the age (23.7 Ma) estimatedin BKFV85.
4. The Oligocene[Eocene(Rupelian to Priabonian)boundary
has been correlated to a level within the upper part of
chronozone C13r (= C13r(0.14)) in the proposed stratotype
Massignano section in the Apennines [Nocchi et al., 1986].
Harland et al. [1990] give a chronogramestimateof 36.4 Ma for
this boundary. More recent assessments
of radioisotopicage
data [e.g., Berggren et al., 1992], however, suggestan age for
the boundarythat is at least 2 m.y. youngeras constrained,for
example, by dates of 33.9 Ma to 34.6 Ma on volcanic ashes
from the lower part of chronozone C13r at Massignano

from earliestEocenemarine depositsin Denmark and the North

Sea Basin, Swisher and Knox [1991] estimatean age of 55 Ma
for the nannofossilNP9/NP10 boundary [see also Berggren et
al., 1992]. This level, which approximatesthe position of the
Paleocene/Eoceneboundary occurs about 2/3 down in chron
C24r (= C24r(0.66)) accordingto Berggrenet al. [1985a, b].
7. The Cretaceous/Paleogene(Maastrichtian to Danian)
boundarylies 3/10 down in chronC29r (= C29r(0.3)) according
to an averageof 5 magnetostratigraphic
sections[Preisingeret
a!., 1986]. Harland et al. [1990] obtained a chronogram
estimateof 66 Ma for the Cretaceous/Paleogene
boundary(based

[Montanari et al., 1988] and a date of 33.4 Ma on oceanic

on 20 dates listed

basementassociatedwith anomaly 13 (= chronC13n) at Ocean

Drilling Project Site 706 [Duncan and Hargraves, 1990]. Odin
et al. [1991] have critically evaluated the bio- and
magnetostratigraphicallywell-controlled radioisotopic dates
from the Paleogenesequencein the Apenninesand basedon
additional analyses, obtain an estimate for the Eocene/
Oligocene boundary of 33.7+0.4 Ma which we use for

Danian), essentiallythe samevalue suggestedby Obradovichet

al. [1986] basedon work on a magnetostratigraphic
sectionin
Red Desert Valley, Alberta. The 66 Ma age estimatewhich we

for the Maastrichtian

and 15 dates for the

adopt
is alsosupported
bylaser
fusion
40Ar/39Ar
dates
on

single crystals of sanidine extracted from the iridium-bearing

lower Z Coal in Montana (C. Swisher and L. Dingus, written
communication,1990; see Berggren et al. [1992] for discussion
calibration.
of apparentlyconflictingdata reportedby Izett et al. [1991]).
5. Subdivisionsof the Eocenedo not provide chronogram
8. In Umbria, the Maastrichtian/Campanianboundary is
estimates suitable for time scale calibration according to
placed in the late part of Chronozone C33n (= C33n(0.15))
Harland et al. [1990], whereasrecent work by Prothero and
[Alvarez et al., 1977; Harland et al., 1990].
For a
Swisher [1992] indicates that the radioisotopic dams from
biostratigraphicallycorrelative level in the Western Interior of
terrestrial deposits in Wyoming [Flynn, 1986] that were the
North America, Obradovich and Cobban [1975] obtained
basisfor a calibrationpoint of 49.5 Ma for the youngerend of
constrainingK-At dateson bentonitesof 74 to 75 Ma (corrected
chron C21n in BKFV85 are likely to be anomalouslyold.
to new decayconstants),which suggestan age of about74.5 Ma
Bryan and Duncan [ 1983] obtaineda K-At date of 46.8+0.5 Ma
for
the boundary(seeBerggrenet al. [1985a, pp. 194-195] for
on biotites from sediments in DSDP
Hole 516F which are
discussionof correlations). Information that supportsthis age
magnetobiostratigraphicallyconstrained to the upper part of
estimate has been provided by Obradovich et al. [1986] who
chron C21n (= C21n(0.33)) [Berggren et al., 1983,1992]. This
report
an40Ar/39Ar
date
of73.4Maforabentonite
near
thetop
calibrationis supportedby datesof 46.2 Ma (K-At) and45.7 Ma
(Rb-Sr) reported by Harris and Fullager [1989] on bentonire of a normal polarity magnetozoneidentified as Chron C33n in
from the lower Castle Hayne Formation of the Ariantic Coastal the San Juan Basin, New Mexico, and by J.D. Obradovich
Plain which is biostratigraphicallyconstrainedto the older part (written communication, 1990) who obtained a date of 75.2 +
0.5 Ma on a bentonite in southwestern Arkansas from the lower
of chron C20r [Berggren et al., 1992].
part of the Globotruncana calcarata zone (the base of this
6. For the Paleocene[Eocene(Ypresian to Thanetian)
boundary, Harland et al. [1990] rejected their chronogram biozone occurs about 1/3 down in Chron C33n in Umbria and is
estimate of 53.4 Ma as too young becauseit is controlled hence approximately 0.5 - 1 m.y. older than the
entirely by glauconites;insteadthey usedan interpolatedage of
Maastrichtian/Campanianboundary which is placed at the G.

75

CALl BRATION

INTERPOLATED

POINT
ANOMALY

AGE

2__5
..

6O

15
3O

_
-

.,,.-

15

200

4OO

60O

8OO

1000

1200

1400

1600

1800

DISTANCE

2000

(Km)

Fig. 27. Ages of magneticanomalies(crosses)as determinedby fitting a cubic splineapproximationfunctionto the

calibrationpoints (inverted triangles).

CANDE AND KENT: NEW GEOMAGNETIC POLARrrY TME SCALE

calcarata/G. tricarinata zonal boundary [Alvarez et al., 1977]).

9. Arthur and Fischer [1977] show the Campanian/Santonian
boundaryat a level several meters below the top of the Gubbio

13,935

GEOMAGNETIC
POLARITY TIME SCALE
LATE CRETACEOUS TO RECENT

Long Normal Zone, the young end of chron C34n (=

C34n(0.0)). On the basis of biostratigraphically controlled

PL ElS TOCEIVE

40Ar/39Ardates
ona bentonite
fromtheWestern
Interior
of

-o

1 '"

Mo

PL.10- L
CEtVE E

North America, Obradovich et al. [1986] confirm an age of

about 84 Ma for the Campanian/Santonianboundary that had
beensuggested
by Obradovichand Cobban [1975] (correctedfor
new decay constants). We estimate an age of 83 Ma for the
stratigraphically younger C34n(0.0) level.

PIACENZIAN
2A--ZANCLEAN3 ... I
MESSINIAN

3A..-

3B-'-__

TORTONIAN 4A'--__

-10
_

Determinationof a New Time Scale

SERRAVALIAN

A natural cubic spline functionwas fit to the nine calibration

age-distance data plus the origin to interpolate the ages of
anomalies. The calibration points and the anomaly ages are
shownin a distanceversusage plot in Figure 27. The new time
scale(CK92; Table 6) is presentedin Figure 28, , th geological
correlationsto anomaliesat the stagelevel, adjustedto the new
derivedages,basedon BKFV85.
In Figure 29 we show an expandedversionof the time scale,
with names for all of the polarity chrons and subchrons. The
occurrencesas presently known of cryptochrons (those tiny
wigglesmodeledas eventsless than 30 ky duration,Table 7) are
indicated by short horizontal lines to the right of the reversal
pattern, reflecting the uncertaintyin their interpretationand the
nonuniformity of coverage.
In Tables 6 and 7 we give the ages of the normal polarity
intervals to a resolution of 1 kyr. This is necessarybecause
severalof the tiny wiggles convert to cryptochronsthat are less
than 10 kyr in duration and consequentlymight be lost if the
agesof the boundingpolarity reversalswere roundedoff to the
nearest10 ky. Between Tables 6 and 7, we recognizea total of
92 normal polarity chrons and subchrons(and of course a like
numberof reversedpolarity intervals)and 54 cryptochronsover
the past 83 m.y.

LANGHIAN

5C-'_

5D---

BURDIGALIAN
5E---

6
6A---

- 20
_

AQUITANIAN

6B.-.

6C ....

CHATTIAN

--

9
_

1ooo

- 30

RUPELIAN 12_...

PRIABONIAN
16

17--_

BARTONIAN

18'-'

-40
_

LUTETIAN

Comparisonto PreviousTime Scales

- 50
_

A comparisonof our new time scale to some previous

geomagnetic polarity time scales is shown in Figure 30.
Differences can be expected as a result of the variety of
calibration data and methodologies that have been used to
constructthe time scales,as well as changesmade to the relative
spacing of the polarity intervals.
Within the Neogene (to chron C6Cn), there are two key
differencesbetweenour time scale and previousversions. First,
since we used the astrochronologiccalibration of 2.60 Ma for
the Matuyama/Gaussboundary that is about 5% older than the
chronogramestimate of 2.48 Ma based on conventional K/At
radiometric dates [Mankinen and Dalrymple, 1979], our time
scale gives proportionately older ages for other geomagnetic
reversalboundariesin the youngerpart of the time scale. There
is already excellent agreement in the few available direct
comparisons between astrochronologic and high precision
radioisotopic age determinations, as outlined above in the
description of calibration points, and we expect that further
refinements in the chronology of Plio-Pleistocene and older
reversalswill be soonemergewith the increasingapplicationof
thesehigh resolutiondating methods.
The secondkey difference in the Neogene is the age of
anomaly 5 (= chron C5n): our time scale gives predicted age
limits (9.592 to 10.834 Ma) that are about0.5 to 1.0 m.y. older
than in virtually all previous time scales (e.g., 8.92 to 10.42
Ma in BKFV85, GTS89; 8.35 to 9.90 Ma in LKC77). We did
not use chron C5n as a calibration point. Moreover, more
widely bracketing age constraints are not that different, for
example, our calibration of C5Bn(0.0) at 14.80 Ma is very

YPRESIAN

24..-

SELANDIAN

26.--

- 60
_

DANIAN

28"-

29-._

- 70

MAASTRICHTIAN

CAMPANIAN

- 80
_

SANTON

IAN

Fig. 28. New geomagneticpolarity time scaleor the Late Cretaceous

and Cenozoic. Correlation of geologic stageswith polarity reversal
pattern is taken from BKFV85. Polarity events shorterthan 30 kyr
durationhave been omittedfrom the reversalpattern.

13,936

CANDE AND KENT: NEW GEOMAGNETICPOLARITYTIME SCALE

similar to the interpolatedage of 14.87 Ma in BKFV85 whereas

the revised age we use for the Gauss/Matuyamacalibrationpoint
accountsfor a only a small part of the shift of citron C5n to an
older age in our time scale (i.e., C5n would be 0.14 m.y.
younger if 2.48 Ma was used instead of 2.60 Ma for C2An
(0.0)). Therefore, the different age we derive for C5n primarily
results from changesin the relative spacingof the anomalies,a
stretching between anomalies 4A and 5 and a relative
contraction between anomalies 5 and 5A. Our predicted age
range for citron C5n agreeswell with the most recent estimates
basedon radioisotopicdating of Icelandic lavas by McDougall
et al. [1984], who give ages of 9.64 Ma and 11.07 Ma for the
boundaries of a normal polarity magnetozonethey correlate
with chron C5n. Age estimates for the base of citron C5n
obtained previously from the Icelandic sections(10.30 Ma by
McDougall et al. [1984] and 10.47 Ma by Saernundsson
et al.
[1980]),
as well
as estimates from East African
magnetostratigraphicsections(10.0 Ma by Tauxe et al. [1985]

and 10.3 Ma by Deino et al. [1990]), are less compatiblewith

our interpolatedage for citronC5n.
In the Paleogene(nominally citronC6C to citronC29), where
we have completely revised the relative spacingsof polarity
intervals,there are many detaileddifferencesbetweenCK92 and
previous time scales. For example, even though the age for
C6Cr(0.0) is very similar (within 4).3 m.y.) in CK92 compared
to GTS89, BKFV85,

TABLE 6. Normal Polarity Intervals

Normal Polarity
Interval, Ma

0.000
0.984

1.757
2.197
2.600
3.127
3.325
4.033
4.265
4.611
4.812
5.7O5
6.078
6.744
6.946

7.153
7.245
7.464
8.047
8.529
9.O69
9.428
9.592
9.777

10.940
11.378
11.852
12.108
12.618
12.718
12.941
13.263
13.674
14.164
14.800
15.038
16.035
16.352
16.583
17.310
18.317
19.083
21.021
21.787
22.166
22.471

0.780
1.049
1.983
2.229
3.054
3.221
3.553
4.134
4.432

Polarity
Chron

Cln
Clr. ln
C2n

LA81 and LKC77, the duration of citron

C6Cr is about a factor of two shorterin CK92 comparedto these

othertime scalesderivedfrom HDHPL68 (-0.7 m.y. versus-1.4
m.y.). A major causeof variation in the Paleogene,however,
stemsfrom the differencesin the agesusedfor the calibrationof
the various time scales. Age estimates of the limits of the
Paleogene (the Oligocene/Miocene at 23.8 Ma and
Cretaceous/Paleogene
at 66 Ma in this paper) may not have
changedby more than -1 m.y. in the decadesinceLA81, but the
ages estimatedfor subdivisionsof the Paleogenehave evolved
considerably(e.g., the Eocene/Oligoceneboundary:38 Ma in
TABLE 6. (continued)

Normal Polarity
Interval, Ma

22.599
22.814
23.357

C2r. ln
C2An.ln

23.678
23.997

C2An.2n
C2An.3n
C3n.ln
C3n.2n
C3n.3n
C3n.4n
C3An.ln
C3An.2n
C3Bn
C3Br. ln
C3Br.2n
C4n.ln
C4n.2n
C4r. ln
C4An
C4Ar. ln
C4Ar.2n
C5n.ln

24.722
24.826
25.482
25.807
25.974
27.004
28.255
28.550
29.373
29.737
30.452
33.050
34.669
35.368
35.716
36.665
37.667
37.988

C5n.2n

38.500

22.760
23.076
23.537
23.800
24.115
24.772
25.171
25.633
25.934
26.533
27.946
28.484
28.716
29.633
30.071
30.915
33.543
34.959
35.554

4.694
5.046
5.946
6.376
6.901
6.981
7.187
7.376
7.892
8.079
8.861
9.149
9.491
9.735
10.834
10.989
11.434
12.000

C5r. ln
C5r.2n
C5An.ln

12.333
12.649
12.764

C5An.2n
C5Ar. ln
C5Ar.2n

41.353
42.629
46.284
48.947
50.646

13.094
13.476
14.059
14.608
14.890
15.162

C5AAn
C5ABn
C5ACn
C5ADn
C5Bn. ln
C5Bn.2n

50.913
52.238
52.641
52.791
55.981
57.800

16.318
16.515
16.755
17.650
18.817

C5Cn. ln
C5Cn.2n
C5Cn.3n
C5Dn
C5En

61.555
63.303
64.911
66.601
68.745

20.162
20.752
21.343
21.877

C6n
C6An.ln
C6An.2n
C6AAn

71.722
72.147
73.517
73.781

71.943
73.g88
73.b84
78.781

83.000

(118.0)

22.263

C6AAr. ln

22.505

C6AAr.2n

39.718

37.534
37.915
38.183
39.639
40.221
41.617
43.868
47.861
49.603
50.812
51.60

53.250
58.1T1
61.951

65.732
68.625

Polarity
Chron

C6Bn. ln
C6Bn.2n
C6Cn. ln
C6Cn.2n
C6Cn.3n
C7n.ln
C7n.2n
C7An
C8n.ln
C8n.2n
C9n
C10n. ln
C10n.2n
Clln. ln
Clln.2n
C12n
C13n
C15n
C16n. ln
C16n.2n
C17n. ln
C17n.2n
C17n.3n
C18n. ln
C18n.2n
C19n
C20n
C21n
C22n
C23n.1n
C23n.2n
C24n. ln
C24n.2n
C24n.3n
C25n
C26n
C27n
C28n
C29n
C30n
C31n
C32n.1n
C32n.2n
C32r. ln
C33n
C34n

CANDE AND KENT: NEW GEOMAGNETIC POLARITY TIME SCALE

LA81, 36.6 Ma in BKFV85, 35.4 Ma in GTS89, and 33.7 Ma in

13,937

EVALUATION: SEAFLOOR SPREADINGRATES IN OTHER OCEANS

this paper; the Paleocene/Eocene

boundary:54.9 Ma in LA81,
57.8 Ma in BKFV85, 56.5 Ma in GTS89, and 55 Ma in this

paper).
For the broadly spaced polarity intervals in the Late
Cretaceous(chrons C30n to C34n), differences are typically
within ~1 m.y. among the most recent time scales (BKFV85,
GTS89, and CK92), but this apparentstability may be more a
reflectionof the sparsityof new data than of well determinedage
calibration

levels.

TABLE 7. CryptochronsFrom C1 to C13 and From C24 to C28

Interval,Ma

0.493
1.201
2.420
8.463
10.060
10.317
10.589
17.860
24.469
25.325
26.328
27.364
27.591
28.091
28.995
29.157
30.251
31.201
31.451
31.825
32.000
32.171
32.432
32.589
32.761
33.261
33.677
33.880
34.174
34.424
53.462
53.604
53.821
53.967
54.170
54.490
54.739
54.953
55.071
55.309
55.611
56.827
57.002
57.160
57.426
57.586
58.756
59.382
59.612
59.821
60.261
60.638
60.931
64.704

Ctyptochron

0.504
1.212
2.441
8.479
10.069
10.342
10.605
17.888
24.480
25.34 1
26.339
27.382
27.609
28.103
29.008
29.164
30.265
31.219
31.460
31.844
32.009
32.180
32.451
32.599
32.771
33.277
33.694
33.888
34.191
34.441
53.471
53.613
53.830
53.976
54.180
54.500
54.749
54.963
55.081
55.319
55.622
56.845
57.020
57.169
57.435
57.604
58.777
59.403
59.622
59.832
60.282
60.648
60.941
64.732

Cln-1
C lr.2r- In
C2r.2r-1
C4r.2r-1
C5 n.2n- 1
C5n.2n-2
C5n.2n-3
C5Dr-1
C6r-1
C7r- 1
C8 n.2n- 1
C9n-1
C9n-2
C9r-1
C10r-1
C10r-2
Cllr-1
C12r-1
C12r-2
C 12r-3
C12r-4
C12r-5
C12r-6
C12r-7
C12r-8
C13n-1

C13r-1
C13r-2
C13r-3
C13r-4
C24r-1
C24r-2
C24r-3
C24r-4
C24r-5
C24r-6
C24r-7
C24r-8
C24r-9
C24r-10
C24r-ll
C25r- 1
C25r-2
C25r-3
C25r-4
C25r-5
C26r-1
C26r-2
C26r-3
C26r-4
C26r-5
C26r-6
C26r-7
C28r-1

The new time scaleindicatesa broadly varyingrate of seafloor

spreading in the South Atlantic (Figure 31, top). The Late
Cretaceouswas characterizedby a rather sharpdecreasefrom the
highest (full) rate of almost 70 mm/yr at around anomalies 33
and 34 time (Campanian) to the lowest rate of less than 30
mrn/yr by around anomaly 27 time (early Paleocene). Over the
Paleogene, spreadingrates first increased to reach a high of
about55 mrn/yr at anomaly15 (latest Eocene),then gradually
decreasedover the Oligocene and early Miocene. Spreading
rates decreasemore rapidly startingin the middle Miocene to the
recentvalue of 32 mm/yr. This systematicpatternof spreading
inferred for the South Atlantic, especially over the Cenozoic
(anomaly 29 to ridge axis), is new and unique to the time scale
derived here, all previous time scales showing a much less
regular variation in spreadingrate over the Cenozoic (Figure
31).
The assumptionof a smoothspreadinghistory for the South
Ariantic in our method of construction,in conjunctionwith the
calibration tiepoints chosen to convert distance into time,
obviously contribute strongly to the regular character of
spreading rate variations with our timescale. While this seafloor spreadinghistory for the South Atlantic seemsreasonable,
less circular assessments
of the geomagneticpolarity time scale
derived from the South Ariantic can be made by determiningthe
implicationsof the new time scale to the spreadinghistoriesof
other ridge systems. In particular, the assumptionof smooth
South Atlantic spreadingcan be testedby calculatingspreading
rates on various other ridges using the new time scale. We
expect to find fewer fluctuationsin spreadingrate on the other
ridges with the new time scale than with previous time scales,
i.e., accelerations should be minimized within the constraints

of the calibration data. This first order test also gives us the
opportunity to examine whether there were globally
synchronousvariationsin spreadingrate, which was one of the
motivating reasonsfor re-examiningthe time scale.
In the sectionbelow, we comparespreadingrate histories
based on our new time scale to the recent time scales of BKFV85

and GTS89 for several different ridge systems, including the

East Pacific Rise at 20S back to anomaly 5A, the Chile Ridge
back to anomaly 5E, the SoutheastIndian Ridge along 130E
back to anomaly 20, the Pacific-Antarctic Ridge south of the
Menard FractureZone back to anomaly 15, the Central Ariantic
north of the Kane Fracture Zone back to anomaly 34, and the
North Pacific along 40N back to anomaly 34.
Method

As in the South Atlantic, the key to examining spreading

rates on other ridges is in the determination of an accurate
synopsis of the magnetic anomaly pattern that takes into
accountlocalized occurrencesof ridge jumps, propagatingrifts
and varying amounts of asymmetrical spreading. Therefore,
whenever possible, we have developed spreading histories
based on (1) an averaging of conjugate profiles or (2) the
calculationof closely spacedfinite rotation poles. In the two
cases where we calculate finite rotation poles (the Southeast
Indian Ridge and the East Pacific Rise), we have used published
rotation poles at intervalsof 5 to 10 m.y. and simply calculated
finite rotation angles that best fit the anomalies in a narrow
corridor. With this method we can account for spreading
asymmetrieswithout having to analyze an entire ridge system,
which is beyond the scopeof this project.
Errors

Errors in spreadingrate are a function of both errors in the

time

scale

and errors

in the estimates

of the widths

of the

anomalies. We assumethat the errors in the time scale (i.e., in

13,938

CANDEAND KENT:NEW GEOMAGNETICPOLARITYTIME SCALE

Chrons

Chrons

(6A_n)6An.2 n
6Ar
1 r.1

lr

lr.2

6AAr

2n

6Bn.2
2r. 1

2r

6Br

2r.2

2An.1

2An.$

6Cn.1

6Cn 6Cn.2

2An

2Ar

6Cr
7n

7n.2

an

8n.2

3n.1
$n.

n
$n. 4

3r

8r
3An
SAn.2

9n

Ar

9r
4n.1

4n
4r

4n.2

4r.2

1On
1Or
11n.1

4An

11n 11n.2 n

4Ar.1

4Ar

11r

4At.2

12n

5n

5n. 2

12r
5r
5r.$

5An. 1

5An 5An.2 n
5Ar.1

13n

5ABn
5ABr

15n

5ACn

15r
5ADn

16n.1

16n

16n.2

17n.1

16r

5Br

5Cn.1

17n

5Cn

17n.; )

5Cr

17n.$

18n.1

18n.2

17r

5Dn
5Dr

18n
5En

5Er

18r

6r

6An

19n

6An. 1

19r

Fig. 29. Expanded

versionof new geomagnetic
polaritytimescaleshowingdetailednomenclature
of chronsandpolarity
events.Namingconvention
is modifiedfromHarlandet al. [1990],asexplained
in text. Tinywigglescorrespond
to polarity
eventsof lessthan30 kyr durationand are shownby shorthorizontallinesto the right of the polaritycolumn.

CANDEANDKENT:NEWGEOMAGNETIC
POLARITY
TIMESCALE

Chrons

Chrons

(27r)

(19r)

28n

20n
28r

29n

20r

29r

30n
21n

31n

21 r

22 n
;31r
22r

23n.1
32n.1

23n

23n.2

32n
32n.2

23r

24n.1

32r

24n
24n.3

24r

33n

25n

25r

26n

26r
33 r

27n

27r

34n

Fig. 29. (continued)

13,939

13,940

CANDEAND KENT:NEW GEOMAGNETIC

POLARfrYTIME SCALE
CK92__

GTS89

BKFV85

LAB1

LKC77

HDHPL68

o
o

:50

4O

AGE

(Ma)

I , I
50

I
.

!
the width of the polarityintervals)are of the samemagnitudeas

00

20

410 i
AGE

60

810

(Me)

the errors in the widths of the anomalies in the South Atlantic

(Table4); theseare typicallyabout7% but varybetween1% and

17% depending
on the anomaly.In Figure31 (top) we showthe
effect of these errors, at the 95% confidence level, on the

anomaly-by-anomaly
estimateof spreadingratesin the South
Atlantic. The fact that the shapeof the SouthAtlanticspreading
rate profi!e is smoothis largely a function of the spline,
whereasthe particularshapeof the curve is dependenton the
particularset of calibrationpointsused.
The errorsin spreadingrate on otherridge systemscan only
be meaningfullyestimatedfor those ridges where we have
calculatedfinite rotation angles. In general,a finite rotation
angle can be estimatedwith an accuracyof 0.05 to 0.1

Fig. 31. Half spreading

ratesin the SouthAtlanticbasedon the new
time scale (top), comparedto three other time scales,including
HDHPL68 (bottom),as evaluatedover roughly2 Ma increments. The
distancesbetween anomaliesare taken from the compositesynthetic
How line data shownin Figure 5. The smooth,broad variationsin

spreading
ratebasedon CK92 are a resultof applyinga splinefunction
to the set of nine calibrationpoMts. The light lines drawnaboveand
below the CK92 profile representthe 95% confidencelimits as
discussedin the text.

East Pacific Rise at 20S

dependingon the quality of the reconstruction.This error

translates
into a wide rangeof absoluteerrorsin spreading
rates

A syntheticspreadingflow line wasconstructed

for the East
Pacific Rise back to anomaly 5A basedon magneticanomaly
spreading
rate;for a fastspreading
ridgetheerroris relatively data between 15S and 25S. The magnetic anomaly picks
small, for a slow spreadingridge the error can be quite large. shown in Figure 32 were digitized and then finite rotation
angleswerecalculatedthatwouldbringtheconjugate
picksback
For both the East Pacific Rise and the SoutheastIndian Ridge we
together.The finite rotationpoleof Minsterand Jordan[1978]
have calculatedthe error in spreadingrate assumingan error of
0.05 in the rotation angle. These error bars are displayedon for the Pacific-Nazcaplate for anomaly3 to presentwasassumed
Figures33 and 36, respectively. For the East Pacific Rise the and only the angle of rotation was calculated. The resulting
errorsare roughly4% for a 1 m.y. intervaland2% for a 2 m.y. finite rotationparametersare given in Table 8.
Using the finite rotationparameters
for anomaliesJ to 5A,
interval, while for the SoutheastIndian Ridge the error is more

depending
on the width of the anomalyintervaland the

typically8% for a 1 m.y. intervaland4% for a 2 m.y. interval.

The total errorin any given spreadingrate is a combinationof
the error due to the estimateof the local anomalywidthsand the
error due to the global estimateof the polarity durations.If we
assumethe two errors are independentthen we can estimatea
total error by taking the squareroot of the sum of the squares.
For the East Pacific Rise, the averageerror is roughly8%, while
for the SoutheastIndian Ridge it is closerto 9%.

spreading
ramswerecalculated
basedon the CK92, GTS89and
BKFV85 time scalesas shownin Figure 33. This figure shows
that,usingGTS89 andBKFV85, therewas an intervalof faster
spreadingrate in the late Neogeneof a similarmagnitudeand

timing as thesetime scalespredictedfor the SouthAtlantic

(Figure 31). With CK92 this intervalof fasterspreadingis
removed,and, if anything,therehas been a gradualdecreasein
spreadingrate sinceanomaly5A.

CANDEANDKENT:NEWGEOMAGNETIC
POLAmTYTMESCALE
-127
-10

-120

13,941

-110

-lOO

-lO

ARRET

.[i::'?i 3

-20

- -20

-27
-127

-120

-27
- 100

-110

Fig.32. Location
ofmagnetic
anomaly
picksontheEastPacific
Riscthatwereused
toconstrain
a setof finiterotation
poles

for thelast12Ma. Theresulting

synthetic
flowlinespreading
historyis shown
by theheavylinesouthof theGarrettFracture
Zone.

TABLE 8. Finite Rotation Parameters: East Pacific Rise

Latitude

Longitude

Angle

Anomaly

56.64
56.64
56.64
56.64
56.64
56.64
56.64

-87.88
-87.88
-87.88
-87.88
-87.88
-87.88
-87.88

18.5
13.0
8.40
5.70
3.60
2.48
1.32

5A
4A
3A
3
2A
2
J

100Ji

3A4A 5A
,
'
CK92.

8O

5A 4A
GTS89
5A

Chile Ridge

Bo

Magneticanomalyspacingsfor the Chile Ridge were

constructed
basedon a long aeromagnetic
profile collectedin

1990 that crossedboth flanksof the ridge(seeFigure8 for

location) [Tebbens et al., 1990].

a_

m lOO

The distances between

% 4A
? 2A

anomalieson the two conjugateprofileswere averagedto

producea composite
profileof magneticanomalyspacings
from

BKFV85

5A

anomaly5E to the ridge axis.

Spreading
ratesfor theChileRidgefromanomaly
5E to the
ridgeaxiswerecalculated
forCK92,GTS89andBKFV85(Figure
34). An intervalof fasterspreading
betweenanomalies5 and
3A (3 in BKFV85) is observedbasedon all three time scales.
Howeverthe magnitudeof the increaseis aboutone-halfto one-

third the amplitudeusingthe CK92 time scalethanusingthe

EPR

6O

10

AGE

20S

15

(Mo)

other two time scales.

Fig. 33. Half spreadingrateson the East Pacific Risc basedon three

different
timescales.Distances
between
anomalies
wereconstrained
by
the syntheticflow line spreading
historyshownin Figure32. Note

Southeast
Indian Ridge

The SoutheastIndian Ridge is characterized

by large
asymmetries in spreading rates.

In order to calculate a

representative
magneticanomalypatternfor this ridge we

that the intervalof fasterspreading

betweenanomalies4A and 3 based

onBKFV85andGTS89is notobserved
based
onCK92(top). Thelight
lines drawn above and below the CK92 curve are an estimateof the error
limits as discussed in the text.

13,942

CANDE AND KENT: NEW GEOMAGC

POLARtrY TIME SCALE

Pacific-AntarcticRidge
5O

5B

CK92

It is difficult to derive a reliable set of anomalyspacingsfor

the Pacific-Antarctic Ridge. The biggest problem is the
sparsityof data on the east flank of the ridge and the lack of
ship tracks that followed along flow lines of spreading. We
have selected an individual track, the Southtow cruise of the R V

:5O

'GTS89

4O

5C 5E

3A
5 _1-'
LU

4o

BKFV85

3O

Washington, that followed a flowline on the west flank of the

ridge southof the Menard FractureZone (Figure37). This cruise
crossedthe ridge 6nly out as far as anomaly3 on the eastflank.
We calculatedanomalyspacingsbasedon an averageof the two
flanks out to anomaly 3, and using just the west flank data
betweenanomaly3 and 15.
Spreadingrate variationsbasedon the Southtowline were
calculatedfor the threetime scalesand (Figure38). The history
of spreadingin the late Neogene was similar to that on the
SoutheastIndian Ridge: basedon CK92, the principalvariation
is an increasein spreadingrate around anomaly 3A; based on
GTS89, the principal changewas aroundanomaly5; basedon
BKFV85, the peak in spreadingwas betweenanomaly3A and 3.
The wide swingsin spreadingrate which apparentlyoccurred
betweenanomalies5D and 12 illustratethe problemof tryingto
interpretspreadingrate variationson a singleprofile from one
flank of a ridge where it is difficult to distinguishvariationsthat
are caused by minor discontinuities in spreading on this
particular flowline from those that are characteristicof the
entire ridge system.

2O

Central

Atlantic

A profile collectedon leg 93 of theRV AtlantisH crossedthe

10

CHIl
0

10

AGE

E' RIDGE'
15

(Ma)

Fig. 34. Half spreadingrateson the Chile Ridgebasedon threedifferent

time scales. The spacingbetween anomalieswas constrainedby the
aeromagneticprofiles, and their extension,shownin Figure 9. Note
that the amplitudeof the pulseof fasterspreading
betweenanomalies5
and 3 basedon BKFV85 and GTS89 is greatlyreducedrelativeto CK92.

analyzed shipboard magnetics data from just east of the

DiscordantZone between 125E and 135E(Figure 35). As on
the EastPacific Rise, we digitizedthe locationof anomalypicks
in a narrow corridor and calculated finite rotation angles for
closely spacedintervals. The rotation poles of Weisselet al.
[1977] were used to provide the latitude and longitudeof the
rotation poles; we simply calculatednew anglesof rotation to
bring the conjugateanomaly picks back together. Based on
these finite rotation parameters (Table 9), we synthesizeda
magneticanomalypatternalong a syntheticflowline at 139E.
Spreading rates along the synthesized flow line were
calculated based on the CK92, GTS89 and BKFV85 time scales

(Figure 36). The pattern of spreadingrate variationsderived

with the CK92 time scaleis considerablysimplerthan with the
other two time scales. We note that a single step-wiseincrease
in rate of roughly25% occurredin the late Neogeneat anomaly
3A time basedon CK92. Accordingto GTS89 therewas a larger
increase(roughly35%) at anomaly5, while BKFV85 displays
an increase at anomaly 5 and 3A and an abrupt decreaseat
anomaly3. Both GTS89 and BKFV85 causethe large swingsin
apparentspreadingrate between anomalies7 and 6A that were
also seen in the South Atlantic.

With the CK92 time scale these

Central

Atlantic

between

the Kane

and Atlantis

fracture

zones

from anomaly 34 on the east flank to anomaly34 on the west

flank (Figure 39) following a flow line. We averaged the
spacingsbetweenanomalieson the two ridge flanks in order to
derive a reliable representation of the magnetic anomaly
spacings.
Spreadingramsfor the CentralAtlantic(Figure40) showlarge
fluctuations between anomalies 4A and 3A with GTS89, between
anomalies

6 and 8 and between

anomalies

18 and 20 with both

GTS89 and BKFV85. These large fluctuationsare diminished

usingCK92 (Figure40). The generalpatternof a periodof slow
spreadingin the latest Cretaceousfollowed by faster spreading
in the Paleoceneand into the early Eoceneis observedbasedon
all three time scales.

North Pacific

We representedthe magneticanomalypatternin the North

Pacific by constructinga compositeprofile near 40N. This
composite profile consisted of (1) two Surveyor-Seamap
profilesacrossthe Juande Fuca Ridge and out to anomaly5E
(profilesIDOE16 and IDOE17 in Figure 14), (2) the averaged
sections we used for constructing the detailed time scale
spacingsbetween anomalies5E and 20, (3) a profile from
Pioneer7103 collectedjust north of the Murray FractureZone
between anomalies20 and 28 (profile P7103-2 in Figure 14),
and 4) a profile from Pioneer7103 from southof the Surveyor
FractureZone betweenanomalies28 and 34 (P7103-1 in Figure
14). For times younger than anomaly20, this compositeline
correspondsto spreadingon the Pacific-Vancouversectionof
the ridge. We have to assumethat spreadingwas symmetrical
becauseonly one half of the spreadingsystemexists,exceptfor
the smallconjugatepart on the Juande Fucaplate.
Spreadingrate variationsin the North Pacific for the last 80

wide swingshavebeenreplacedby a muchmoregradualdecrease

in spreading rates between anomalies 7 and 6C. Between
anomalies7 and 12 the CK92 based spreadingrates are also
smoother. Prior to anomaly 12 time there were wide swings

scales,are shownin Figure 41. It is apparentthat with CK92

there are actually larger swings in spreadingrate than with
BKFV85 and GTS89, particularlybetweenanomalies6C and 13,

based on all three time scales.

near anomalies 17 and 20, and between anomalies 32 and 33.

Ma based on the CK92, as well as the BKFV85 and GTS89 time

CANDE
ANDKENT:NEWGEOMAGNETIC
POLARITY
TIMESCALE

12.0

130

3-33

21-24

13,943

140

i,a:.
.

."

-r./.,

-4O

-4O

4
e,i.,=...

'

ee.... e

-5O

e--

-='-'

e"e--e eee-4--e---e--e
3A

ea--e
e- 3-' -

--

-5O

-6O

-6O

.18 '

120
150
140
Fig.
35.Location
ofmagnetic
anomaly
picks
between
Australia
and
Antarctica
used
tocalculate
aset
offinite
rotation
poles

for
thelast
45Ma.
Thefinite
rotation
poles
were
used
todetermine
asynthetic
flowlinespreading
history
justwest
ofthe
George
V Fracture
Zone(heavyline).

13,944

CANDEAND KENT: NEW GEOMAGNETIC

POLARtrYTIME SCALE

TABLE 9. Finite Rotation Parameters: Southeast Indian Ridge

Latitude

10.3
10.3
10.3
10.7
11.1
11.3
11.9
12.5
12.85
13.83
14.25
14.167
14.041
13.878
13.8
13.617
13.373
13.005
12.580
11.241

Longitude

34.75
34.75
34.75
34.6
34.6
34.5
34.4
34.4
34.12
33.52
33.25
33.501
33.882
34.369
34.6
34.686
34.801
34.973
35.172
35.793

Angle

Anomaly

23.7
23.3
22.6
22.0
21.3
20.85
20.1
18.6
17.86
16.3
15.48
14.80
13.85
12.27
11.50
11.07
10.56
9.81
9.06
7.37

20
19
18
17
16
15
13
12
11
9
8
7
6C
6A
6
.
5D
5C
5B
5A

9.7

36.5

6.07

9.7

36.5

5.47

4A

9.7
9.7
9.7

36.5
36.5
36.5
36.5
36.5
36.5

4.74
3.83
2.63

4
3A
3
2A
2
J

6. Thus the sharp swings observed in all of the ridges except

the North Pacific usingHDHPL68 and derivativetime scaleslike
BKFV85 and GTS89 were the resultof taking the North Pacific
pattern and inserting the sharp drop at anomaly 6C into the
South Atlantic pattern. The sharp swings aroundanomaly 20
may be due to a similareffect.
TECTONIC IMPLICATIONS

The new time scale provides an opportunityto investigate

changes in plate motion on a finer scale than previously
possible. We have re-examined earlier observationsof global
changesin spreadingrate basedon HDHPL68 andits derivatives
with our independentlyderivedtime scale. As is apparentfrom
the discussionin the previoussection,the previoustime scales
gave the misleading impression that there was a globally
synchronous"pulse" of faster spreadingin the late Neogene,
startingaroundanomaly5 time and endingaroundanomaly3A
or 3 time [e.g., Vogt, 1986]. With the new time scalewe see a
much simpler pattern of spreadingrate changes. The Central
and South Atlantic are generally seen to be slowly varying
while, in contrast,more frequentabruptchangesoccurredin the
spreading rates of ridges in the Pacific and Indian Oceans.
Althoughwe do not see evidencefor proposedglobal pulsesin
spreadingrate, we can documentmajor, regionally synchronous
changesin plate motion.
A clear example of a regional plate reorganizationthat is
better understood

with the new time scale is the circum-Pacific

event that occurredin the late Neogene. Cox and Engebretson

[1985] and Pollitz [1986] proposedthat there was a changein
9.7
1.12
9.7
.62
the absolutemotion of the Pacific plate at about5 Ma, whereas
Harbert and Cox [1989] proposedthat it occurredat a younger
time, at about3.5 Ma. With previoustime scalesit was difficult
to relate the changesin spreadingrate on the various ridge
Discussion
systemswith this changein absoluteplate motion. We now see
that there was an increase in spreadingrate on the PacificWe observethat for all of the ridge systemsexaminedexcept Antarctic Ridge at around anomaly 3A time (Figure 42). The
the North Pacific (and perhapsthe South Pacific for which the change in spreading rate on the Pacific-Antarctic Ridge
data base is poor), the spreadingrate variationsare considerably correspondsto a dramaticchangein the characterof the geoid
smootherusing CK92 than usingBKFV85 or GTS89. We see anomaly over the southernmostfracture zones on this ridge
that the new time scale either eliminatesor greatly reduces1) system as imaged in the Geosat altimeter data [Haxby et al.,
the "pulse"of faster spreadingobservedbetweenanomalies5 1992]. Simultaneously,we see a sharp increasein spreading
and 3 (or 3A dependingon the time scale)in the SouthAtlantic, rate on the SoutheastIndian Ridge at anomaly3A time (Figure
the EPR and on the Chile Ridge, (2) the sharpswingsbetween 42) which also correspondsto a time of a 20 clockwiserotation
anomalies 7 and 6 in the South Atlantic, the Southeast Indian
in spreading direction on the SoutheastIndian Ridge [e.g.,
Ridge and in the Central Atlantic, and (3) the large swings Munschyet al., 1992]. These spreadingchangesmay correlate
around anomaly 20 in the South Atlantic and in the Central with the change in Pacific absolute plate motion, although,
Atlantic. We also note that betweenanomaly6C and the ridge using the new time scale,we would suggesta slightly older age
axis the SoutheastIndian Ridge spreadingrate was constant of 5.6 Ma.
Other circum-Pacificeventsthat occurredat nearly the same
except for a step at anomaly 3A. These observationsstrongly
supportour initial assumptionthat seafloor spreadingrates in time (anomaly 3A) include the initial rifting of the Lau Basin
the South Atlantic were smoothly varying without the sudden which wasdatedat 5.6 Ma by drilling on ODP Leg 135 [Leg 135
changesobservedin the Pacific. This is a somewhatdifferent Scientific Party, 1992], the rifting of Baja California [Larson,
pictureof spreadingin the SouthAtlantic than that presentedby 1972], and a decreasein spreadingrate on the Chile Ridge
Aubry et al. [1988], who found distinct breaks in the South (Figure 34).
A second example of a significant plate motion
Atlantic pattern. We suggestthat thesebreakscan be attributed
to irregularitiesin the individual profiles used by Aubry et al. reorganizationthat can be viewed in a differentperspectivewith
[1988].
the new time scaleare the circum-PacificeventsaroundAnomaly
It is clear that there were several large, rapid changesin
6C time, near the Oligocene/Mioceneboundary. This time
apparentspreadingrate in the North Pacific, in particular,the interval is associatedwith the breakupof the Farallon plate into
sharpdrops at anomaly 20 time and at anomaly6C time. We the Nazca and Cocosplates [Hey, 1977; Lonsdaleand Klitgord,
can thus identify the sourceof some of the "artifacts" that are 1978]. The old time scales predicted a slowdown in North
presentin the spacingsin HDHPL68 and that have been handed Pacific spreadingrates and erratic swingsin spreadingrates on
down from time scaleto time scalethroughto GTS89. Heirtzler virtually all of the other ridge systems. With the new time
et al. [1968] used the North Pacific to define the details of the scale, we see a precipitous drop in Pacific-Vancouver (North
reversal pattern and interpolated this pattern into the South Pacific) spreadingrates(Figure42) that correlatesin time with a
Atlantic framework. In some parts of the time scale the large change in the position of the rotation pole [Wilson,
interpolationpoints were quite far apart; e.g., betweenanomaly 1988] and a small, but distinct, drop in spreadingrate on the
5 and 7 there was only a single interpolationpoint at anomaly SoutheastIndian Ridge (Figure 42). This suggeststhat the
9.7

1.68

CANDEANDKENT:NEWGEOMAGNETIC
POLARITY
TIMESCALE

4O

I.- [:::::::l
3A

12I-I16
I

5E

6C

13,945

CK92

3O

2O

40

6A 6C

12

''

LU

3o

36 GTS89
2O

3 3A
I

LU

40

6A 6C
5

12

BKFV85

30

2O

SOUTHEAST

10

INDIAN

RIDGE

20

30
AGE

40

(Me)

Fig. 36. Half spreading

rateson the Southeast
IndianRidgefor thelast45 Ma basedon threedifferenttime scales.The
spacings
between
anomalies
aretakenfromthesynthetic
flowlinespreading
history
in Figure35. Notethatthefluctuations
in spreading
ratesinceanomaly
12 aregreatlyreduced
relativeto CK92. ThelightlinesdrawnaboveandbelowtheCK92
curve are an estimate of the error limits as discussed in the text.

-1;50

-120

10
-45

-5o

-50

-150

-120

-110

-55

Fig.37. Location
of theSouthtow
profilein theSouthPacific
usedto constrain
thespreading
ratehistory
of thePacificAntarctic Ridge.

13,946

CANDEANDKENT:NEWGEOMAGNEC
POLARtrYTIMESCALE

SoutheastIndian Ridge systemis againrespondingto a major

changein Pacific plate motion.
A major puzzle of Pacific tectonicsis that while a large

some 10 m.y. older than the bend. We do howeversee a large

dropin Pacific-Vancouver
spreading
rateat 43 Ma (Anomaly20)
which also correlateswith the time of increasedspreadingrates

changein the absolutemotionof the Pacificplate apparently on the Southeast Indian Ridge [Cande and Mutter, 1982].
occurred at 43 Ma as inferred from the bend in the Hawaiian-

Emperorchain,whichin BKFV85 corresponded

to anomaly19,
therewasno apparentchangeat thattime in the Pacific-Farallon
(Vancouver)relativemotion;the major changein the direction
of Pacific-Farallonspreadingoccurredaroundanomaly24 time,
which was dated at 55 Ma [e.g., Atwater, 1989]. With the new
time scale,anomaly24 has becomeabout2 m.y. younger,still

4O

r 40
a_ .30

20-

PACI F/C-ANTARCT/C
51S

10-

R/DGE

20

AGE

.30

(Ma)

40

Althoughit is temptingto correlatethesechangesin spreading

rate to the bend in the Hawaiian-Emperorchain, it has to be

emphasizedthat the spreadingrate historyof the North Pacific

is basedon a single compositeprofile, and that the conjugate,
Farallonplate side of the spreadinghistoryis long gone.
The North Pacific can now be viewed as a systemwith large
swingsin spreadingrate relative to the smoothspreadingrate
history of the South Atlantic. This irregular spreadingrate
history (at least as recordedon our single compositeprofile)
may reflect the changing configurationof subduetingslabs
aroundmuch of the perimeterof the plate. The irregularitiesin
Pacific-Vancouver spreadingrates appear to increase as the
Vancouverplate diminishesin size, again apparentlyreflecting
the effect of the varying subduetingslab. One unexplained
aspectis that the Pacific-Vancouverspreadingrate changedso
drasticallyat anomaly6C time, when the primaryPacificbasin
event is supposedlythe breakupof the Farallon plate. It is
surprising that there is so much linkage of events on the
Farallonplate to eventson the Pacific-Vancouverridge system
and even, apparenfiy,on the SoutheastIndian Ridge.
The Late Cretaceousremains a time of poorly constrained
spreadingrates. With our calibrationpointsit is clearthat there
mustbe a sharpdrop in spreadingratesin the SouthArianticin
the late Cretaceous. However, becauseof the sharpnessof the
drop, and the relative sparsityof calibrationpoints,the actual
timingof the dropis not well constrained.It may haveoccurred
graduallybetweenanomalies33 and 30, as we proposehere, or
it may have occurredmore abruptlyat aroundanomaly29, near
the Cretaceous/Paleogene
boundary. This is a critical time
interval to resolve with more numerous and precise age
calibration data becauseof the large changesin spreadingrate
that must have occurred in the Indian Ocean [e.g., Schlich,
1975; Patfiat, 1983].

SUMMARY AND CONCLUSIONS

Fig. 38. Fluctuations
in thehalf spreading
rateon the Pacific-Antarctic
ridgebasedon threedifferenttime scales.anomalyspacings
aretaken
from the Southtowprofile shown in Figure 37. Fluctuationsin
We have constructeda new geomagneticpolarity time scale
spreading
rate relativeto all threetime scalesshowsdifficultiesof for the late Cretaceousand Cenozoic based on an analysisof
constraining
spreading
ratehistorybasedon a singleprofile.
magneticprofilesfrom the world'smajoroceanicridgesystems.
-60
:55

55

50

5O

25

25

20
-60

2O

Fig. 39. Locationof the A119321profilefromthe CentralAtlantic.

CANDEAND KENT:NEW GEOMAGNETIC

POLARtrYTIME SCALE
33

22 24

34

2O

CK92

8 111318
,0

5A5C
4A

15

13,947

6A

22

GTs89

6013

3A ,.

'

20

a:: lO

BKFV85

S.jl
A51A
C
8 17
CENTt?AL

ATLANTIC

27N

20

40

60
AGE

80

(Me)

Fig.40. Fluctuations
in halfspreading
ratein theCentralAtlanticrelative
to threedifferent
timescales.Anomaly
spacings
weredetermined
by averaging
thedistances
between
anomalies
onbothflanksof theMid-Atlantic
ridgeontheA11-93
profile.
Note the smootherspreading
ratehistoryoverthe last 50 Ma relativeto CK92.

80

This project was initiated becauseof our suspicionthat there

were errors in the relative widths of the polarity intervalsin
HDHPL68 andthat theseerrorsresultedin misinterpretations
of

10

6C

tectonic events.

117 Il

Illhfi I1=,
40 4A.
,SAC

sequence.

Thereareseveralsignificant
changes
in thespacings
andages
of the polarity intervals. For example,the derived age of
anomaly 5 is about 0.5 m.y. older than in most other time

scales but is now in agreement with the results of the most

recent radioisotopic ages of basaltic flows on Iceland. The

C 3o

7o

40

-z

30

13

34

5Ac

, , 1 33

7 lO
, ,

'
17

In addition, we felt that there was now a

sufficientincreasein the amountof magneticprofile data to

justify a reappraisal of the complete magnetic anomaly

effect of this change is that the global "pulse" of faster

spreadingthat appearedto have occurredbetweenanomalies5
and3 hasbeeneliminated.In addition,the relativespacings
of
reversalsbetweenanomalies5 and 3 have beenchanged. As a
result, a late Neogene plate motion changeon the Southeast
IndianRidgeis now seento be temporally
relatedto a spreading
rate changeon the Pacific-AntarcticRidge and also with other
tectoniceventsthatoccuredaroundanomaly3A time throughout

II

the Pacific.

Othersignificant
changes
to thereversalsequence
weremade
betweenanomalies7 and 6. Erratic swingsin spreadingrate
thathad beenpreviouslyinterpreted
in mostoceans,exceptthe
NorthPacific,havebeenreplacedwith muchsmoother
changes,
or no change,while the North Pacificnow showsa precipitous

4o "5
2o

-- '
' ' 33BKFV85

drop in spreadingrate at anomaly6C time.

NORTHPACIFIC
COMPOSITE

lO

PROFILE

80

AGE

(Me)

Fig. 41. Half spreadingratesin the North Pacificrelativeto threetime

scales.

Distances

between

anomalies

were

constrained

from

compositeprofile. Note the larger fluctuationsin spreadingrate

relativeto CK92, particularlyaroundanomaly6C, betweenanomalies
10 and 7, and between anomalies 20 and 17.

The ageof reversalsbetweenanomalies

7 and24 are2-3 m.y.
younger than in most previous time scales, reflecting the
revisedagesof calibrationpointsin the Paleocene-Oligocene.
The effect of these changesis that spreadingrates in several
oceansappearto have a largebut gradualincreasestartingin the
mid-Eoceneand culminatingin the early Oligocene.
The largestuncertainties
in our new time scaleare probably
for the anomalyspacingsin the Late Cretaceousand early
Cenozoic. This correspondsto the time interval of the most
rapid change in spreadingrate (and rapid migration of the
rotationpoles)in the SouthAtlantic and, consequently,
a time

13,948

CANDEANDKENT:NEW GEOMAGNETIC
POLARflYTIME SCALE
,

EPR
......

20S

North

Pacific

South Pacific

9O

EPR

Chile Ridge
Southeast
South

..........

8O

Indian

Ocean

Atlantic

Central

Atlantic

7O

60
Chile
--1

"'

I
I

5O

South

I- -L__

Pacific

I--

I__

I
J

--

'---i

Pacific

I
i

40 -

North

...., i

r-i

F' i

Southeast
Indian

SE.I [-' I i.-L_!._i

i i i-i

:50

r-

'
_._,.__.
,

20 --

South
e...
...... ,%

Atlantic
....

lO -"'

eeee

......

I.. ,. ,.. ''t... .,

ee

eeeee

'

''' ''''

Central

Atlantic

I,,

I,

10

15

20

25

30

35

,I

AGE (Mo)
Fig. 42. Half spreadingrates in the late Paleogeneand Neogene on several different ridges. Note the synchronous
fluctuationson the SoutheastIndian, Pacific-Antarctic(SouthPacific) and Chile ridgesat 5.6 Ma.

when there is the most potential for error in defining the South
Atlantic spreadingpatternand its age calibration.
One interestingimplication of our study is that spreading
rates in the North Pacific now appear to have been quite
variable, starting in the mid-Eocene and continuingthroughthe
Oligocene. We speculatethat as the Farallon and Vancouver
platesbecamesmaller,their motionwas moreandmoreeffected
by small variations in the configurationof the subductingslabs
around their eastern perimeters. Spreading in the South
Atlantic, where there is little interaction with subduction zones,

is probably driven primarily by more gradually evolving ridge

push. This interoceanvariation in driving mechanismsjustifies
our dependenceon anomaly spacingsin the South Atlantic, as
opposedto platesin the Pacific, to constructthe new time scale.
We have also identified

the source of some of the errors that

are apparentin HDHPL68 and its descendents.HDHPL68 was a

blend of two profiles: the Verna 20 profile from the South
Atlantic and a profile from the North Pacific. The data from the
North Pacific were insertedinto the South Atlantic spacingsat
variable and often quite large intervals. The erratic swings in
spreadingrate betweenanomalies7 and 6, for example,can now
be attributed to the use of the spacingsexclusively from the
North Pacific over this long time interval in earlier time scales.
Finally, we have identified many additional anomalies that

may represent reversals of the global geomagneticfield, for

example,betweenanomalies3A and 4A. Most importantly,we
observe an essentially continuous pattern of small scale
anomaliesbetween anomalies24 and 27 that appear to be an
"Earth-filtered" record of short period (2 to 20 kyr) intensity
variations of the dipole field. We believe that this type of
dipole field behavior,previouslyrecognizedwithin anomaly5
and between anomalies 12 and 13, may have characterizedthe
geomagneticdynamo throughoutthe Cenozoic. The present
resultsprovide motivation for a continuedstudy of small scale
anomalieson high-resolutionmarine magneticprofiles.
Further refinement of the geomagnetictime scale in general
and its tectonic and geomagneticconsequencesawait more
detailedanalysesof magneticanomalypatternsalong flow lines
on the ocean ridges, and more precise and closely spacedage
calibration points.
APPENDIX: NOMENCLATURE

The polarity chronnomenclaturewe use is similar to that of

Tauxe et al. [1983] and Harland et al. [1990], with some

modifications. Like Hariand et al. [1990], we refer to the

longest intervals of predominantly one polarity by the
correspondinganomaly number followed by the suffix n for

CANDE AND KENT: NEW GEOMAGNETICPOLARITY

normal polarity, or r for the preceding reversed polarity

interval.

When

these chrons

are subdivided

into

shorter

polarity intervals, we refer to them as subchronsand identify

them by appending,from youngestto oldest, a 0.1, 0.2, etc., to
the primaxychronname, and addingan n for a normal polarity
interval, or an r for a reversedinterval. For example, the three
normal polarity intervals composinganomaly 6C (chron C6Cn)
are called subchrons C6Cn. ln, C6Cn.2n, and C6Cn.3n, whereas
Harland et al. [1990] refer to them as chrons C6C.ln, C6C.2n,
and C6C.3n. Similarly we refer to the reversed interval

preceding(older than) subchronC6Cn.ln as subchronC6Cn.lr,

whereas Harland et al. [1990] call this interval chron C6C.lr.
We use this form of nomenclatureso that we can name every
chron and subchronwithout resortingto odd constructions. For
example, there are polarity subdivisions of both chrons C4n
and C4r, which we designate as SubchronsC4n. ln, C4n.2n,
C4r. lr and C4r.2r (Figure 29), and which Harland et al. [1990]

would designateC4.1n, C4.2n, C4.1r, and C4.2r, respectively.

SCALE

13,949

Acknowledgments. This project benefited from discussionswith

many of our colleagues,includingTanya Atwater, William Berggren,
William Lowrie, Peter Molnar, John Obradovich,Philippe Patriat,
Walter Pitman, Hans Schouten,and Carl Swisher. We especially
thank Bill Ryan and Ken Miller, who reviewedthe manuscript,Bill
Haxby, who provided valuable assistancein calculating cubic
splines, and the JGR reviewers, who provided constructive
criticisms. The aeromag-netics data from the Chile Ridge were
collectedduringa joint Naval ResearchLab-LDGO-Chileprojectwith
Skip Kovacs,Juan Carlos Parra, JohnLaBrecque,and SCC as coinvestigators. Data from the Indian Ocean, used to define the fine
details of the reversalpattern betweenanomalies24 and 27, were
obtainedfrom Roland Schlichat EOPG Strasbourgas part of a joint
US-French Indian Ocean data compilation project. SCC's

participationin boththe Chile Ridge andIndian Oceanprojectswas

fundedby the NationalScienceFoundation.Betty Batchelderassisted
with the drafting, and Bel Hautau and Diana Saaby contributedto the
preparationof the manuscript. This work was supportedby grants
OCE88-11339
and OCE91-04447
Foundation.
LDGO contribution 4951.

of

the

National

Science

With the Harland et al. [1990] nomenclature, the reversed

intervals preceding C4.1n and C4.2n cannot be named C4.1r

and C4.2r without confusing them with the subchronswithin
chron C4.1r, thus forcing the use of the awkward designation
C4.1nr and C4.2nr; with our nomenclature, we refer to these
intervals

as C4n. lr and C4n.2r.

For more precisecorrelation,the fractionalpositionwithin a

chron or subchronis referred to by the equivalent decimal
number appended,within parentheses,to the chron or subchron
name, following the approachof LaBrecque et al. [1983]. As
examples, the younger end of chron C29n is C29n(0.0) or
C29(0.0) (= C29n(y) in Harland et al. [1990]), the older end of
chron C29n (=
C29n(o) in Harland et al. [1990]) is
conveniently designatedas C29r(0.0) since it is equivalent to
the younger end of Chron C29r, whereas a level within chron
C29r and 3/10 from its youngerend is referred to as C29r(0.3).
We use the designation-1,-2, etc., following the primary
chron or the subchrondesignationto denote apparently very
short polarity intervals corresponding to the tiny wiggles
which, upon calibration, convert to durations of less than 30
kyr. In view of their uncertainorigin, we refer to theseglobally
mapped geomagneticfeatures as cryptochrons. Thus the tiny
wiggles between anomalies12 and 13 (within chron C12r) are
called from youngest to oldest cryptochronsC12r-1, C12r-2,
etc. This differs from Harland et al. [1990] who used a duration

of 100 kyr or less to distinguishthe class of shortestpolarity

intervals, a cut-off value which often fragments the chron
heirarchy in a manner inconsistentwith the magnetic anomaly
identifications. For example, the four positive anomalies
composing anomaly 3 (which correspond to the Cochiti,
Nunivak, Sidufjall, and Thvera subchronsof the Late Cenozoic
K-At geomagnetic reversal time scale [Mankinen and
Dalrymple, 1979]), are designatedby Harland et al. [1990],
from youngestto oldest, as C3.1n, C3.2n, C3.2r-ln and C3.3n,
while we refer to them as C3n. ln, C3n.2n, C3n.3n and C3n.4n,
respectively.
We believe that the 30 kyr cut-off is more realistic in
separatingthe anomalies that are well characterizedand most
probably reflect true geomagneticpolarity reversalsfrom the
smaller scale anomalies (i.e., tiny wiggles) that are generally
lessuniformly well documentedand whoseorigin may be due to
paleointensity variations or incomplete reversals of the
geomagneticfield. A cryptochroncan be elevatedto the status
of a subchron if it is demonstrated that the tiny wiggle
correspondsto a pair of geomagnetic polarity reversals, as
appearsto be the case for the Cobb Mountain subchronat about
1.1 Ma [Mankinen et al., 1978; Mankinen and Gromme, 1982;

Clement and Kent, 1987]. In suchcaseswe also add a polarity

suffix to the name, hence the Cobb Mountain

designated Clr.2r-ln.

subchron is

Alvarez,W., M.A. Arthur,A.G. Fischer,W. Lowfie, G. Napoleone,and

I. Premoli-Silva,Upper Cretaceous'-Paleocene
magneticstratigraphy
at Gubbio, Italy, V, Type sectionfor the late Cretaceous-Paleocene
geomagneticreversal time scale, Geol. Soc. Am. Bull., 88, 383-389,
1977.

Andreieft, P., H. Bellon, and D. Westercamp, Chronometrie et

stratigraphie comparee des edifices volcaniques et formations
sedimentaires
de la Martinique (Antilles francaises),Bull. Bur. Rech.
Geol. Min., 4, 335-346, 1976.

Arthur, M.A., and A.G. Fischer,Upper Cretaceous-Paleocene

magnetic
stratigraphy at Gubbio, Italy, I, Lithostratigraphy and
sedimentology,Geol. Soc. Am. Bull., 88, 367-371, 1977.
Atwater, T., Plate tectonichistory of the NortheastPacific and Western
North America, in The Geology of North America, vol. N, The
Eastern Pacific Ocean and Hawaii, edited by E.L. Winterer, D.M.
Hussongand R.W. Decker, Geological Society of America, Boulder,
Colo., 1989.

Atwater,T., and J. D. Mudie, Derailednear-bottomgeophysicalstudyof

the Gorda Rise, J. Geophys.Res., 78, 8665-8686, 1973.
Atwater, T., and J. Severinghaus,Tectonic maps of the Northeast
Pacific, in The Geology of North America, vol. N, The Eastern
Pacific Oceanand Hawaii, editedby E.L. Winterer,D.M. Hussongand
R.W. Decker, GeologicalSocietyof America, Boulder,Colo., 1989.
Aubry, M-P., W.A. Berggren, D.V. Kent, J.J. Flynn, K.D. Klitgord,
J.D. Obradovich,and D.R. Prothero, Paleogenegeochronology:an
integratedapproach,Paleoceanography,3, 707-742, 1988.
Baksi, A., B. Houghton, M. McWilliams, H. Tanaka, and G. Turner,
What is the age of the Brunhes-MatuyamaPolarity Transition?,Eos,
Trans. AGU, 72, 135, 1991a.
Baksi, A., K.A. Hoffman, and M. McWilliams, Preliminary resultsof

40Ar/39/Ar
dating
studies
directed
to testing
theaccuracy
of the
geomagneticpolaritytime scale(GPTS) at 2-5 Ma, Eos, Trans. AGU,
72, 135, 1991b.

Baksi,
A.K.,V. Hsu,
M.O.McWilliams,
andE.Farcar,
40Ar/39Ar
dating
of the Brunhes-Matuyamageomagneticfield reversal,Science,256,
Barker, P.F., The history of ridge-crestoffset at the Falkland-Agulhas
FractureZone from a small-circlegeophysical
profile, GeophysJ. R.
Astron. Soc., 59, 131-145, 1979.
Berggren, W.A., N. Hamilton, D.A. Johnson,C. Pujol, W. Weiss, P.
Cepek,and A.M. Gombos,Jr., Magneto-biostratigraphy
of Deep Sea
Drilling Project Leg 172, Sites 515-518, Rio Grande Rise (South
Atlantic), Init. Rep. Deep Sea Drill. Proj., 72, 939-948, 1983.
Berggren,W.A., D. V. Kent, andJ. J. Flynn, Paleogenegeochronology
and chronostratigraphy,The chronology of the geological record,
Mem. Geol. Soc. Am., 10, 141-195, 1985a.

Berggren, W.A., D. V. Kent, J. J. Flynn, and J.A. Van Couvering,

Cenozoic geochronology, Geol. Soc. Am. Bull., 96, 1407-1418,
1985b.

Berggren, W.A., D.V. Kent, J.D. Obradovich, and C.C. Swisher IlI,
Toward a revised Paleogenegeochronology,in Eocene-Oligocene
Climatic and Biotic Evolution, edited by D.R. Prothero and W.A.
Berggren,29-45, PrincetonUniv. Press,Princeton,N.J., 1992 .

13,950

CANDE AND KENT: NEW GEOMAGNEC POLARITY TIME SCALE

Blakely,R.J., Geomagnetic
reversalsand crustalspreading
ratesduring
the Miocene, J. Geophys.Res., 79, 2979-2985, 1974.
Bryan,N.B., andR. A. Duncan,Age andprovenance
of elastichorizons

Hey, R. N., Tectonicevolutionof the Cocos-Nazcaspreadingcenter,

Geol, Soc. Am. Bull., 88, 1414-1420, 1977.

Hilgen,F.J.,Astronomical
calibration
of Gaussto Matuyamasapropels
in the Mediterranean
andimplicationfor the geomagnetic
polarity

from Hole 516F, Init. Rep.Deep SeaDrill. Proj., 72, 475-477, 1983.
Cande, S.C., and D.V. Kent, Ultra-high resolutionmarine magnetic
time scale,Earth Planet. Sci. Lett., 104, 226-244, 1991a.
anomalyprofiles: a recordof continuouspalco-intensityvariations?, Hilgen,F.J., Extension
of the astronomically
calibrated(polarity)time
J. Geophys.Res., in press, 1992.
scaleto the Miocene/ Plioceneboundary,Earth Planet. Sci. Lett.,
107, 349-368, 1991b.
Cande, S.C., and Y. Kristoffersen,Late Cretaceousmagneticanomalies
in the North Atlantic, Earth Planet. Sci. Lett., 35, 215-224, 1977.

Cande, S.C., and J. L. LaBrecque, Behaviour of the Earth's

palcomagneticfield from small scale marine magneticanomalies,
Nature, 247, 26-28, 1974.
Cande, S.C., and J.C. Mutter,

A revised identification of the oldest

seafloorspreadinganomaliesbetweenAustraliaand Antarctica,Earth
Planet. Sci. Lett., 58, 151-160, 1982.

Cande, S., J.L. LaBrecque,and W.F. Haxby, Plate kinematicsof the

South Atlantic: Chron C34 to present,J. Geophys.Res., 93, 13,47913,492,

1988.

Cande, S.C., J.L. LaBrecque, R.L. Larson, W.C. Pitman HI, X.

Golovchenko,and W.F. Haxby, Magneticlineationsof the world's
oceanbasins,map with text, Am. Assoc.Petrol. Geol., Tulsa, Okla.,
1989.

Carbotte,S., S.M. Welch, and K.C. Macdonald,Spreadingrates,rift

propagation,
and fracturezoneoffsethistoriesduringthe past5 m.y.
on the Mid-Atlantic Ridge, 25-2730'S and 31-3430'S, Mar.
Geophys.Res., 13, 51-80, 1991.

Champion,D.E., G.B. Dalrymple,and M.A. Kuntz, Radiometricand

palcomagneticevidencefor the Emperorreversedpolarity event at
0.46 +_0.05 m.y. in basalt lava flows from the easternSnake River
Plain, Idaho, Geophys.Res. Lett., 8, 1055-1058, 1981.
Champion,D.E., M.A. Lanphere,and M.E. Kuntz, Evidencefor a new
geomagneticreversalfrom lava flows in Idaho: discussionof short
polarity reversalsin the Bmnhesand late Matuyamapolarity chrons,
J. Geophys.Res., 93, 11,667-11,680, 1988.
Clement, B.M., and D.V. Kent, Short polarity intervals within the
Matuyama: Transitionalfield recordsfrom hydraulicpistoncore sites
in the North Atlantic, Earth Planet. Sci. Lett., 81,253-264,

1987.

Cox, A., and D. Engebretson,

Changein motionof the Pacificplateat 5
Ma, Nature, 313, 472-474, 1985.

Deino,
A., L. Tauxe,
M. Monaghan,
andR. Drake,
40Ar/39Ar
age
calibration of the litho- and palcomagneticstratigra-phiesof the
Ngorora Formation,Kenya, J. Geol., 98, 567-587, 1990.

Duncan,
R.A.,andR.B.Hargraves,
40Ar/39Argeochronology
of
basementrocks from the MascarenePlateau,the ChagosBank, and
the Maldives Ridge, Proc. OceanDrill. Program, Sci. Results,115,
43-51,

1990.

Flynn, J.J., Correlation and geochronologyof Middle Eocene strata

from the western United States, Palaeogeogr. Palaeoclimatol.
Palaeoecol., 55, 335-406, 1986.

Gromme, C.S., and R.L. Hay, Geomagneticpolarity epochs:Age and

duration of the Olduvai normal polarity event, Earth Planet. Sci.
Lett., 10, 179-185, 1971.

Haq, B.U., J. Hardenbol, and P.R. Vail, Mesozoic and Cenozoic

chronostratigaphy
and cycles of sea-levelchange,Spec. Publ., Soc.
Econ. Paleontol. Mineral.,

42, 71-108, 1988.

Harbert, W., and A. Cox, Late Neogenemotion of the Pacific plate,J.

Geophys.Res., 94, 3052-3064, 1989.
Haftand, W. B., A.V. Cox, P.G. Llewellyn, C.A.G. Pickton, A.G.
Smith, and R. Walters,A GeologicTime Scale, 131 pp., Cambridge
University Press,Cambridge,New York, 1982.
Haftand, W.B., R. Armstrong,A.Cox, L. Craig, A. Smith, and D. Smith,
A GeologicTime Scale1989, 263 pp., CambridgeUniversityPress,
New York, 1990.

Harris, N.B., and P.D. Fullagar,Comparisonof Rb-Sr and K-At datesof

middle Eocene bentonitc and glauconite, southeasternAtlantic
Coastal Plain, Geol. Soc. Am. Bull., 101, 573-577, 1989.
Haxby, W.F., C.A. Raymond,S.C. Cande,W.B.F. Ryan, S. Tebbens,S.
O'Hara, D. Mueller, B. O'Brien, and M. Wilkinson, Preliminary

resultsof a Hydrosweep,gravity and magneticssurveyof a PacificAntarctic Ridge fracture zone: PaRR

2, Crustal Structure,Eos Trans.
AGU, 73, 278, 1992.
Heirtzler, J.R., G.O. Dickson, E.M. Herron, W.C. Pitman III, and X.

LePichon, Marine magneticanomalies,geomag-neticfield reversals,

and motionsof the oceanfloor and continents,J. Geophys.Res., 73,
2119-2136,

1968.

Izett,G.A.,G.B.Dalrymple,
andG.B.Snee,40Ar/39Ar
ageofthe
Cretaceous-Tertiaryboundary tektites from Haiti, Science, 252,
1539-1542, 1991.

Kent, D.V., and F. M. Gradstein,A Jurassicto recentchronology,

in
The Geology of North America, vol. M, The WesternNorth Atlantic

Region, edited by P.R. Vogt and B.E. Tucholke, pp. 45-50,

GeologicalSociety of America, Boulder, Colo., 1986.
Klitgord, K.D. and H. Schouten, Plate kinematics of the central
Atlantic, in The Geology of North America, vol. M, The Western
North Atlantic Region,editedby P.R. Vogt and B.E. Tucholke,351377, GeologicalSocietyof America,Boulder,Colo., 1986.

Klitgord,K.D., S. P. Heustis,J.D.Mudie,andR. L. Parker, An analysis

of near-bottommagneticanomalies:Sea floor spreadingand the
magnetizedlayer, Geophys.J. R. Astron. Soc.,43, 387-424, 1975.

LaBreeque,J.L., K.J. Hsu, M.F. Carman,Jr., A-M. Karpoff, J.A,

McKenzie, S.F. Percival, Jr., N.P. Petersen, K.A. Pisciottto, E.

Schreiber,L. Tauxe,P. Tucker,H.J. WeisseRR,

andR. Wright, DSDP
Leg 73: Contributionsto Paleogenestratigraphyin nomenclature,
chronologyand sedimentation
rates,Palaeogeogr.,Palaeoclimatol.,
Palaeoecol., 42, 91-125, 1983.

LaBrecque,
J.L., D.V. Kent,andS.C.Cande,Revisedmagnetic
polarity
time scalefor the Late Cretaceous
and Cenozoictime, Geolo_y,5,
330-335,

1977.

Larson, R.L., Bathymetry,magneticanomalies,and plate tectonic

history of the mouth of the Gulf of Califomia, Geol. Soc. Amer.
Bull., 83, 3345-3360, 1972.
Larson, R.L., and W.C. Pitman Ill, Worldwide correlation of Mesozoic

magneticanomalies,and its implications,Bull. Geol. Soc. Am., 83,

3645-3662,

1972.

Leg 135 ScientificParty,Evolutionof backare basins:ODP Leg 135,

Lau Basin, Eos, Trans. AGU, 73, 241, 1992.

Lonsdale,P., and K.D. Klitgord, Stmctureand tectonichistoryof the

eastern Panama basin, Geol. Soc. Am. Bull., 89, 981-999, 1978.

Lowtie, W., and W. Alvarez, UpperCretaceous-Paleocene

stratigraphy
at Gubbio,Italy, III, UpperCretaceous
magneticstratigraphy,
Geol.
Soc. Am. Bull., 88, 374-377, 1977.

Lowrie,W., andW. Alvarez,One hundredmillionyearsof geomagnetic

polarity history, Geology, 9, 392-397, 1981.
Mankinen,E.A., and G. B. Dalrymple,Revisedgeomagnetic
polarity
time scalefor the interval 0-5 m.y.B.P.,J. Geophys.Res., 84, 615626, 1979.

Mankinen,E.A., and C.S. Gromme,Palcomagnetic

datafrom the Coso
Range, California, and the current status of the Cobb Mountain
normal geomagnetic event, Geophys. Res. Lett., 9, 1279-1282,
1982.

Mankinen, E.A., J.M. Donnelly, and C.S. Gromme, Geomagnetic

polarity event recordedat 1.1 m.y.B.P. on Cobb Mountain, Clear
Lake volcanicfield, Califomia, Geology,6, 653-656, 1978.
McDougall,
I., L. Kristjansson, and K. Saemundsson,
Magnetostratigraphy
and geochronology
of northwestIceland,J.
Geophys.Res., 89, 7029-7060, 1984.
Miller, K.G., M.J. Kahn, M-P. Aubry, W.A. Berggren,D.V. Kent, and
A. Melillo, Oligocene to Miocene bio-, magneto-, and isotope
stratigraphyof the western North Atlantic, Geology, 13, 257-261,
1985.

Minster,J.B., and T.H. Jordan,Present-day

plate motions,J. Geophys.
Res., 83, 5331-5334, 1978.
Montanari, A., A.L. Deino, R.E. Drake, B.D. Turrin, D.J. DePaola, G.S.

Odin, G.H. Curtis,W. Alvafez,and D. Bice, Radioisotopic

datingof
the Eocene-Oligocene
boundaryin the pelagic sequences
of the
northeasternApennines,in The Eocene-Oligocene
Boundaryin the
Marche-UmbriaBasin (Italy), editedby I. Premoli-Silva,R. Cocioni,
and A. Montanari, pp. 195-208, International Subcommission
Paleogene Stratigraphy, Eocene/OligoceneBoundary Meeting,
Ancona,Oct. 1987, SpecialPublication,1988.
Munschy, M., J. Dyment, M.O. Boulanger,D. Boulanger,J.D. Tissot,
R. Schlich,andM.F. Coffin, Breakupand seafloorspreading
between

CANDEAND KENT:NEW GEOMAGNE'I'IC

POLARITYTIME SCALE
the Kerguelen Plateau-Labuan Basin and the Broken RidgeDiamantina Zone, Proc Ocean Drill. Program , Sci. Results,120,
931-944,

1992.

Ness, G., S. Levi, and R. Couch,Marine magneticanomalytimescales

for the Cenozoic and Late Cretaceous:A precis, critique, and
synthesis,Rev. Geophys.,18, 735-770, 1980.
Nocchi, M., G. Parisi, P. Monaco, S. Monechi, M. Mandile, G.

Napoleone,M. Ripepe, M. Orlando, I. Premoli-Silva, and D.M.

Brice, The Eocene-Oligoceneboundary in the Urnbrian pelagic
regression,in Terminal Eocene Events, Dev. in Palaeontol. and
Stratigr., vol. 9, editedby C. Pomerol,and I. Premoli-Silva,25-40,

13,951

Swisher, C.C., and R.W.O'B. Knox, The age of the Paleocene/Eocene

Boundary:
40Ar/39Ar
dating
ofthelower
partofNP10,
NoahSea
Basin and Denmark,paperpresentedat the InternationalGeological
Correlation Project 308: Paleocene/EoceneBoundary, Brussels
meeting, Dec. 2-6, 1991.

Swisher,
C.C.,andD.R.Prothero,
Single-crystal
40Ar/39Ar
dating
of
the Eocene-Oligocenetransition in Noah America, Science, 249,
760-762,

1990.

Talwani, M., C.C. Windisch and M.G. Langseth,Jr., 1971, Reykjanes

Ridgecrest:a detailedgeophysical
study,J. Geophys.Res., 76, 473517.

Elsevier, New York, 1986.

Obradovich, J.D., and W.A. Cobban, A time scale for the Late
Cretaceous of the western interior of Noah America, Geol. Assoc.

Tauxe, L., P. Tucker, N.P. Petersen and J.L. LaBrecque, The

magnetostratigraphy of Leg 73 sediments, Palaeogeogr.,

Can. Spec.Pap. 13, 31-54, 1975.

Obradovich,J.D., J.F. Sutter, and M.J. Kunk, Magnetic polarity chron
tie pointsfor the Cretaceous
and early Teaiary, Terra Cognita,6,

Tauxe, L., M. Monaghan, R. Drake, G. Curtis, and H. Staudigel,

Palcomagnetismof Miocene East African Rift sedimentsand the
calibrationof the geomagneticreversaltime scale,J. Geophys.Res.,

140, 1986.
Odin, G.S., A. Montanari, A. Deino, R. Drake, P.G. Guise, H. Kreuzer,
and D.C. Rex, Reliability of volcano-sedimentary
biotite ages across

the Eocene-Oligoceneboundary (Apennines,Italy), Chem. Geol.,

Isotope Geosci. Sect., 86, 203-224, 1991.
Patriat, P., Reconstitutionde l'evolution du systemede dorsalesde
l'ocean Indien par les methodesde la Cinematiquedes Plaques,
Doctoral d'Etat, Universitc Paris 6, 1983.

Pitman, W.C., III, and J.R. Heirtzler, Magnetic anomalies over the
Pacific-AntarcticRidge, Science,154, 1164-1171, 1966.

Pollitz, F., Pliocenechangein Pacificplatemotion,Nature, 320, 738741, 1986.

Preisinger,A., E. Zobetz, A.J. Gratz, R. Lahodynsky,M. Becke, H.J.

Mauritsch,G. Eder, F. Grass,F. Rogl, H. Stradner,and R. Surenian,
The Cretaceous/Teaiary boundary in the Gosau Basin, Austria,
Nature, 322, 794-799, 1986.
Prothero, D.R., and C.C. Swisher III, Magnetostratigraphy and
geochronologyof the terrestrial Eocene-Oligocenetransition in
Noah America, in Eocene-OligoceneClimatic and Biotic Evolution,
editedby D.R. Protheroand W.A. Berggren,46-73, PrincetonUniv.
Press, Princeton, N.J., 1992
Rea, D.K., and R.J. Blakely, Short-wavelengthmagnetic anomaliesin

Palaeoclimatol. Palaeoecol., 42, 65-90, 1983.

90, 4639-4646, 1985.

Tauxe, L., A.D. Deino, A.K. Behrensmeyer,

and R. Potts,Pinningdown
the Brunhes/Matuyama and upper Jaramillo boundaries: A
reconciliationof orbital and isotopictime scales,Earth Planet. Sci.
Lett., 109, 561-572, 1991.
Tebbens,S.F., S.C. Cande, J. Kovacs, and J.L. Labrecque,Neogene
Antarctic-Nazcaplate motions,Eos Trans.AGU, 71, 1641, 1990.
Tsuchi, R., Y. Takayanagi and K. Shibata,Neogene bio-eventsin the
JapaneseIslands, in Neogene of Japan - Its Biostratigraphyand
Chronology,editedby R. Tsuchi,pp. 15-32, KurofunePrintingCo.,
Shizuoka, 1981.

Vine, F.J.,andD.H. Matthews,Magnetic anomaliesover oceanicridges,

Nature, 199, 947-949, 1963.

Vogt, P.R., Plate kinematicsduringthe last 20 m.y. and the problemof

"present"motions,in The Geologyof North America,vol. M, The
Western North Atlantic Region, edited by P.R. Vogt and B.E.
Tucholke, pp. 405-425, GeologicalSociety of America, Boulder,
Colo., 1986.

Walker, R.C., P.C. Manega, R.L. Hay, R.E. Drake, and G.H. Curtis,

Laser-fusion
40Ar/39Ar
dating
ofBed1,Olduvai
Gorge,
Tanzania,
Nature, 354, 145-149, 1991.

Weissel,J. K., D.E. Hayes, and E.M. Herron, Plate tectonicssynthesis:

a regionof rapidseafloorspreading,
Nature, 255, 126-128,1975.
The displacements
betweenAustralia,New Zealand,and Antarctica
Ryan, W.B.F., Stratigraphyof late Quaternarysedimentsin the eastern
Mediterranean, in The Mediterranean Sea: A Natural Sedimentation

Laboratory, edited by D.J. Stanley,Dowden, Hutchinsonand Ross,

Stroudsburg,Pa., 1972.
Saemundsson,
K., L. Kristjansson,I. McDougall, and N.D. Watkins, KAr dating, geological and palcomagneticstudy of a 5-km lava
successionin noahern Iceland, J. Geophys.Res., 85, 3628-3646,

since the Late Cretaceous,Mar. Geol., 25, 231-277, 1977.

Wilson, D.S., Tectonichistoryof the Juande Fucaridge over the last 40
million years, J. Geophys.Res., 33, 11863-11876, 1988.
Wilson, D.S. and R.N. Hey, The Galapagosaxial magneticanomaly:
evidencefor the emperoreventwithin the Brunhesandfor a two-layer
magneticsource,Geophys.Res.Lett., 8, 1051-1054, 1981.

1980.

Schlich,R., Structureet age de l'oceanIndien occidental,Mem. Hors

Ser. Soc.Geol. Fr., no. 6, 103 pp., 1975.
Schouten,H., and K. McCamy, Filteringmarinemagneticanomalies,J.
Geophys.Res., 77, 7089-7099, 1972.
Shackleton, N.J., A. Berger, and W.R. Peltier, An alternative
astronomical

calibration

of the lower Pleistocene timescale based on

ODP Site 667, Trans. R. Soc. EdinburghEarth Sci., 81, 251-261,

1990.

S.C. Cande,ScrippsInstitutionof Oceanography,

La Jolla,CA 92091.
D. V. Kent, Lamont-DohertyGeologicalObservatoryof Columbia
University, Palisades,NY 10964.
(ReceivedOctober3, 1991;
revisedMay 11, 1992;
acceptedMay 27, 1992.)