Stacey F.D., Davis P.M. Physics of the Earth

Подождите немного. Документ загружается.

//FS2/CUP/3-PAGINATION/SDE/2-PROOFS/3B2/9780521873628C25.3D

–

427

– [417–437] 13.3.2008 10:45AM

possible. The Kent and Smethurst evidence is a

strong bias to shallow magnetic inclinations,

which would be consistent with a random distri-

bution of continents if there were such a strong

octupole field. The alternative interpretation is a

clustering of the sampled continents at low lati-

tudes, which Pesonen et al. (2003) found to be the

case in a series of Proterozoic continental recon-

structions. There is no evidence that the field was

so fundamentally different in the early period

that an octupole field much stronger than the

dipole field at core level is plausible. Rather, we

would expect the dipole field to have been rela-

tively stronger when the inner core was small. So,

the question raised by the Pesonen et al. (2003)

analysis is: what reason might there be for con-

tinents to cluster around the equator? If we had

an answer it would make an interesting topic for

Chapter 12.

Over-riding the question of the dipole/quad-

rupole or multipole structure of the field is its

axial character. By Curie’s principle of symme-

try, which we refer to also in Sections 17.9 and

24.5, when averaged over a suitable time, the

magnetic field must be symmetrical about the

rotation axis, unless there is some permanent

asymmetry in its causes. Since we are not

aware of such an asymmetry, and it seems

unlikely that there is one, we accept the axial

principle, which is central to paleomagnetism

and to global tectonics (Section 25.6).

25.4 Geomagnetic reversals

The discovery of thermoremanent magnetiza-

tion (TRM), induced in pottery, bricks and lavas

as they cooled, dates from at least the early nine-

teenth century. Certainly it was well known in

1906 when the French physicist B. Brunhes

found not only a lava flow but also an adjacent

clay that had been baked by the lava, in which

the remanent magnetism was almost precisely

opposite to the present direction of the field.

Brunhes drew the conclusion that the Earth’s

field must have reversed, but his discovery was

so far in advance of dynamo theory that there

was no way that it could be appreciated fully at

the time. Further similar observations in the

1920s by P. L. Mercanton and M. Matuyama also

attracted little attention until much later.

By the late 1940s, many more reversely mag-

netized rocks were being discovered, but the

concept of a reversed field was not accepted

without question and alternative explanations

were sought. L. N

eel, who had developed the

theory of ferrimagnetism (see Fig. 25.1), under-

took an exhaustive theoretical study of mecha-

nisms by which magnetic remanence in

minerals could, in principle, be self-reversing.

Not all of N

eel’s mechanisms were really plausi-

ble, but in 1952 a lava from Mt Haruna, in Japan,

was found to acquire reversed TRM in laboratory

experiments. An intensive period of theoretical

and laboratory studies established that the self

reversal was due to ionic rearrangement in solid

solutions of ilmenite (FeTiO

) and hematite

(Fe

) (Ishikawa and Syono, 1963). The disor-

dered, high temperature state converts to an

ordered arrangement of ions at low tempera-

tures. Ions that move to new lattice sites are

magnetically aligned by interactions with their

neighbours and develop a reversed remanence

exceeding the original normal one. The reversed

remanence is a property only of an intermediate,

partially ordered state and is not observed in

either the disordered or fully ordered states. It

is a rare phenomenon, but the fact that it occurs

allowed expressions of doubt about geomagnetic

reversals to persist for several years.

Studies dedicated to the reversal problem

eventually made the case for field reversals unas-

sailable. There are four principal observations.

(i) Wilson (1962) found numerous baked con-

tact rocks, of the type first recognized by

Brunhes, and showed that in the over-

whelming majority of cases their rema-

nence directions coincided with those of

the lavas that heated them and generally

not with the directions in neighbouring

unheated rock. Later statistics with larger

numbers have reinforced this conclusion.

(ii) For rocks up to a few million years old,

which can be sufficiently accurately dated,

the dates of normal and reversed rocks coin-

cide on different continents and for differ-

ent rock types (e.g. lavas and sediments). By

25.4 GEOMAGNETIC REVERSALS 427

//FS2/CUP/3-PAGINATION/SDE/2-PROOFS/3B2/9780521873628C25.3D

–

428

– [417–437] 13.3.2008 10:45AM

convention, magnetizations parallel to the

present field are referred to as ‘normal’,

although it is clear that there is nothing

abnormal about reversed magnetization.

(iii) The detailed process of reversal has been

traced in both rapid sequences of lava

flows and in deep sea sediments.

(iv) Linear magnetic anomalies observed at sea

(as in Fig. 12.10) are identified with stripes

of normal and reversely magnetized basalt,

parallel to the ocean ridges where they were

formed (Fig. 12.9). For the few million years

of dating that is sufficiently accurate for a

precise check, the sequence of reversals

required to explain the anomalies coincides

with the magnetic polarities of dated igne-

ous rocks, assuming a more or less steady

spreading of the sea floor away from the

ridges. This is consistent with paleontolog-

ical estimates of the age of the ocean floor

(Fig. 12.8).

Dynamo theory readily accommodates rever-

sals, because a field of either sign can be

maintained equally effectively by any particular

pattern of motion or body forces. Equation (24.36)

is unaffected by reversing the sign of B. All that is

needed is an instability that triggers reversals. The

secular variation demonstrates that there is no

steady, stable state of the dynamo, but reversals

do not appear to be just extremes of the normal

secular variation, although the two phenomena

are presumably related. Gubbins (1994) reviewed

the basic physics of this problem. Growth, decay

and drift of the non-dipole field and of the equa-

torial dipole are continuous processes, and the

axial dipole also fluctuates in strength, but rever-

sals of the axial dipole occur in 5000 years, or

perhaps less in some cases, compared with the

average interval between them, hundreds of

thousands of years. This pattern of behaviour is

suggestive of a chaotic system switching irregu-

larly between two quasi-stable states. However,

there are extreme variations in the frequency of

reversals that require an explanation.

We can examine details of the record of rever-

sals for clues to the dynamo instability that

causes them. There are two semi-independent

aspects to the problem, the field pattern during

a reversal, and the statistics of the intervals

between reversals. There is wider general agree-

ment on the statistical behaviour. The normal

and reversed states occur equally often, with no

difference between their average durations, as

would be expected for a purely random process

that is the same for both polarities. If we suppose

that reversals are independent random events,

with a fixed average rate of occurrence, then the

probability p(t) of an interval of constant polarity

with duration t decreases exponentially with t

(in the same way as the probability of survival

of a radioactive nucleus). This is the Poisson

distribution

pðtÞ/e

t=



; (25:6)

where



t is the average interval of fixed polarity

between reversals and is the same for both polar-

ities. However,



t is not constant with time, but

has varied slowly from about 2 10

years to

more than 10

years. Merrill et al. (1996) consi-

dered the possibility that a reversal might be

followed by a period of inhibition to further

reversals, requiring replacement of Eq. (25.6) by

a gamma function. However, they noted that,

with successive improved reversal chronologies,

the departure from Eq. (25.6) diminished. It

appears likely that there is no inhibition and

that the statistical indication of it arises from

short polarity events that are missed in the

record (but see Problem 25.5).

Figure 25.6 shows the reversal record for the

last six million years, for which the dating of

reversals is most precise, and therefore the

agreement between data from different conti-

nents, and between sediments and igneous

rocks, is most securely observed. Back to about

100 million years ago the framework of the

record is obtained from the sea floor magnetic

anomalies. For earlier times, without the sea

floor record, the sequence is less certain,

although it is evident that the reversing behav-

iour was similar. Extended periods with one

dominant polarity are referred to as chrons,

with recent ones named after early pioneers in

geomagnetism. Briefer polarity intervals (sub-

chrons) are named after the sites where rocks

that record them were first found. Even shorter

428 ROCK MAGNETISM AND PALEOMAGNETISM

//FS2/CUP/3-PAGINATION/SDE/2-PROOFS/3B2/9780521873628C25.3D

–

429

– [417–437] 13.3.2008 10:45AM

events, not represented in the figure, occur and

there are also incomplete or aborted reversals,

termed excursions, that may be too brief to indi-

cate that a full 1808 polarity change occurred, or,

if it did so at one site, whether that was apparent

elsewhere. The sequence of reversals, with irreg-

ular long and short polarity intervals, can be

seen in sufficient detail in sediment cores to

allow dates to be correlated with greater preci-

sion than the determination of absolute ages.

The use of the reversal record for stratigraphic

correlation has become an important adjunct to

paleontological and isotopic methods (Opdyke

and Channell, 1996).

Detailed observations of field changes during

a reversal require a continuous record, such as

that in Fig. 25.7. This was obtained from a marine

sediment core that had been deposited rapidly

enough to minimize direction averaging. It

shows several features that are common to rever-

sal records, most significantly the diminished

field strength for a few thousand years during

the reversal and the progressive decrease for

5.894

Age

(Ma)

Miocene Pliocene Pleistocene

Epoch Polarity

chrons

Brunhes

0.780

Matuyama

Jaramillo (N)

Olduvai (N)

Réunion (N)

Kaena (R)

Mammoth (R)

Cochiti (N)

Nunivak (N)

Sidufjall (N)

Thvera (N)

2.581

Gauss

3.580

Gilbert

5.894

5.230

4.980

4.890

4.800

4.620

4.480

4.290

4.180

3.580

3.330

3.220

3.110

3.040

2.581

2.150

2.140

1.950

1.770

1.070

0.990

0.780

Polarity

subchrons

FI G U R E 25.6 Geomagnetic polarity record for the last

six million years, as plotted by Merrill et al. (1996) from

data by Cande and Kent (1995). Periods of normal

polarity are marked black and reversed polarity white.

Numbers give dates of boundaries in millions of years.

–30

–60

–90

360

270

180

0.1

0.01

1234

9.0

0 567891011

9.5 10.0

Depth (metres)

Relative a

e (thousands of years)

Intensity (Am

–1

)

Declination (degrees)

Inclination (degrees)

FI G U R E 25.7 Detailed record of a reversal from a

rapidly deposited deep-sea sediment core. Samples

were ‘cleaned’ in a 10

2

T alternating field before

measurement. This reversal marks the lower boundary

of the Jaramillo polarity event, 1.07 million years ago

(see Fig. 25.6). Note that time runs from right to left.

Figure redrawn after Opdyke et al. (1973).

25.4 GEOMAGNETIC REVERSALS 429

//FS2/CUP/3-PAGINATION/SDE/2-PROOFS/3B2/9780521873628C25.3D

–

430

– [417–437] 13.3.2008 10:45AM

several thousand years before there were signifi-

cant direction changes. The inclination record

shows repeated strong fluctuations, but this is

probably not very significant because, with the

weakened field, the non-dipole field was more

prominent, causing direction changes that can

be identified only with a particular locality and

not with the polarity of the field as a whole.

Measurements of paleo-intensity during

numerous reversals, such as that in Fig. 25.7,

agree that the field strength is reduced by a fac-

tor of 3 to 10 and, as in this example, the reduced

intensity is generally more prolonged than the

period of rapid direction change. Since the

dipole field normally dominates the observed

field only by a similar factor, the obvious con-

clusion is that the axial dipole dies away, and

redevelops in the opposite direction, with no

accompanying enhancement of the equatorial

dipole, or of the non-dipole field. An alternative,

but disputed, suggestion is that the equatorial

dipole remains strong enough for the field to

retain its dipole character during a reversal, so

that the magnetic north pole crosses the equator

at an identified longitude, instead of disappear-

ing from one hemisphere and reappearing in the

other. The particular interest in this question

arises from reports that the pole not only retains

its identity, but that, during repeated reversals, it

follows preferred paths from pole to pole, either

through the Americas or 1808 away, through

East Asia and Western Australia. Laj et al. (1992)

summarized the results of several authors who

reached this conclusion. The interpretation is

not entirely straightforward and, in reviewing

the evidence, Merrill and McFadden (1999) ques-

tioned the significance of the preferred paths.

We take the position that the effect is real and

that the explanation is a mantle control on mag-

netic flux at the top of the core, for reasons that

follow.

Figure 25.8 shows the representation by

McFadden and Merrill (1995) of the longitudes

of the reported equator crossings by the postu-

lated reversing dipole. Note that it is the path of

the geomagnetic north pole that is plotted, so

that this is a polar effect for which the opposite

polarities are not equivalent. Gubbins (2003)

drew attention to the fact that the preferred

paths provide the most direct indication of man-

tle control on core behaviour and presented fur-

ther evidence. He plotted a graph of the square of

the vertical component of the present field, aver-

aged over all latitudes, as a function of longitude,

and showed that it is highly correlated with a

similar graph of shear wave speed at the base of

the mantle. A strong vertical component (of

either sign) is seen where the mantle appears

cool. Gubbins’s point is that cool mantle enhan-

ces the flux of core heat into it and the resulting

cooled core material is carried down convec-

tively. By the frozen flux principle it carries the

field with it, towards the downwelling, and the

surface expression is a locally strong vertical

field. The striking feature of the Gubbins graphs

is that there are two peaks, at longitudes coinci-

ding with the preferred pole paths during rever-

sals. This partially resolves what has been a

paradox in the reversals story: the preferred

paths can be explained simply as an artifact of

mantle control that concentrates the vertical

field component at selected longitudes.

However, a crucial question remains. The

reported reversal paths show a preference for the

–180

–60

–30

–150 –120 –90 –60 –30

Reversal equator crossin

0 30 60 90 120 150 180

FI G U R E 25.8 Longitudes of

equator crossings of poles during

polarity transitions observed by

sedimentary cores, as plotted by

McFadden and Merrill (1995).

430 ROCK MAGNETISM AND PALEOMAGNETISM

//FS2/CUP/3-PAGINATION/SDE/2-PROOFS/3B2/9780521873628C25.3D

–

431

– [417–437] 13.3.2008 10:45AM

western path (through the Americas). Assuming

this to be statistically significant, since the path

of the magnetic north pole is plotted, conditions

at the core–mantle boundary must distinguish

between the two polarities. It is possible, in prin-

ciple, for chemical differences to generate ther-

moelectric or galvanic currents that would bias

the dynamo, although we believe the core–

mantle boundary to be too nearly isothermal to

admit the thermoelectric alternative as likely.

The only serious possibility appears to be a

chemical battery effect, varying with time as

the material at the base of the mantle is replaced.

This allows the suggestion that the seismic veloc-

ity variations at the base of the mantle, used in

the analysis by Gubbins referred to above, have

an origin at least partly in chemical differences.

While this is just possible, we need to question

the statistical significance of the polarity bias

that it aims to explain.

Reversals occur sufficiently often to allow

observation of a statistically significant variation

in their rate of occurrence (Fig. 25.9). The peak

rate is at least five per million years but, as the

figure shows, there was an extended period of

constant (normal) polarity (disallowing possible

unobserved brief events) from about 118 million

years to 84 million years ago (the Cretaceous

superchron). An even longer period of reversed

polarity (the Permo-Carboniferous superchron)

appears to have extended from 312 to 256

million years ago, but is not confirmed by the

sea floor anomaly record because no ocean floor

has survived from that time. This figure docu-

ments the idea, discussed by McFadden et al.

(1991), that reversals are most frequent when

the symmetric harmonics of the field, referred

to in Sections 24.3 and 25.3, are strongest, rela-

tive to the antisymmetric harmonics. They noted

that the scatter of virtual poles caused by sym-

metric components of the present field, S

,is

independent of latitude, , but that the scatter

by antisymmetric components, S

, is propor-

tional to . Spherical harmonics are orthogonal

functions so that the total scatter, S, is given by

¼ S

þ S

¼ðS

=Þ



þ S

: (25:7)

By assuming this to be true also in the remote

past, McFadden et al. separated the antisymmet-

ric and symmetric components of the scatter, as

in Fig. 25.9. Since we must suppose that the

scatter is proportional to strengths of the com-

ponents, this gives a measure of their relative

strengths. The positive correlation of S

(six aver-

aged data points) with the rate of reversals (solid

line) is clear in the top half of the figure. The

inverse correlation with the antisymmetric field

is shown in the bottom half.

While we grope for an understanding of

the reversal mechanism, and hypotheses

abound, we can sift the plausible from the

implausible to narrow the range. Mantle control

0.2

/ φ

(degees)

0.4

40 60 80 100

Age (Ma)

150 200

40 60 80 100 150 200

Reversals/millions

ears

FI G U R E 25.9 (Top) Variation of reversal

rate for the last 165 million years (jagged

line – right-hand scale), with the latitude-

independent component of the scatter of

pole positions due to secular variation

during periods between reversals.

(Bottom) Latitude-dependent

component of the secular variation (see

Eq. (25.7)). Based on McFadden et al.

(1991) and redrawn from a figure by

Merrill et al. (1996).

25.4 GEOMAGNETIC REVERSALS 431

//FS2/CUP/3-PAGINATION/SDE/2-PROOFS/3B2/9780521873628C25.3D

–

432

– [417–437] 13.3.2008 10:45AM

is sufficiently strongly indicated to consider the

three obvious ways in which it may influence

core motions: thermal, topographic and electro-

magnetic. They are not necessarily independent,

but the extreme variability of the reversal rate

implies a variation in core–mantle conditions

that is localized and cannot reasonably be

global. For example, the efficiency argument in

Section 22.7 requires that any variation in the

total core-to-mantle heat flux be reflected in

dynamo power. The evidence is conflicting.

Shaw and Sherwood (1991) reported no correla-

tion between field strength and the rate of rever-

sals, but Tauxe (2006) found field intensity and

polarity interval duration to be positively corre-

lated. Figure 25.9 indicates that the symmetric/

antisymmetric ratio is diagnostic; it is possibly

the most direct clue that we have. The reversal

mechanism seen in the numerical dynamo

model of Takahashi et al. (2005) may also offer a

lead. In this model, concentrated radial flux

patches appeared at low latitudes and their

migration to high latitudes was followed by

field reversal in a manner qualitatively similar

to the equatorward migration of sunspots in the

11-year cycle of solar field reversals. Radial flux

at the equator is characteristic of symmetric field

components (axial quadrupole and equatorial

dipole), so, if the core–mantle boundary struc-

ture is such as to promote the development of

flux patches of the kind seen in the Takahashi

et al. model, it could be reflected in the strength

of the symmetric harmonic terms.

It may be possible to merge the flux patch

idea with a theory by Olson (1983), which re-

quires core motions to be driven by two sources

of buoyancy (or negative buoyancy), one originat-

ing at the inner core boundary and the other at

the core–mantle boundary. The idea is that the

two sources of buoyancy generate motions of

opposite helicities in the rotating fluid. The

sign of helicity can be thought of as the direction

of rotation as one follows a spiral path and is

opposite in the two hemispheres for each source

of buoyancy. The heterogeneity of the D

layer at

the base of the mantle allows the possibility of

local variations in the heat flux conducted into it

and, therefore, in the generation of localized

negative thermal buoyancy at the top of the core.

Several, but not all, studies have indicated that

the inner core inhibits reversals, because, being

solid, the only way that the field in it can change

is by diffusion. Its electromagnetic relaxation

time is about 1600 years, which is shorter than

the time for a reversal, but sufficient to add some

inertia to the dynamo. If this is significant then

we might expect reversals to have been more

frequent when the inner core was smaller, but

the detailed reversal history is too short to give an

indication of the behaviour 2 billion years or so

ago and, in any case, such an effect would prob-

ably not be observable against the background of

variations that occur anyway. There is a better

chance of seeing an effect of the change in inner

core size from measurements of the intensity of

the early field, because a smaller inner core con-

tributes less of the two most important compo-

nents of core energy – compositional separation

and latent heat (see Section 22.7).

25.5 Paleointensity – the strength of

the ancient field

There is a long history of paleointensity mea-

surements, but reliability is a serious problem

for several reasons. The now classical method

was developed by Thellier and Thellier (1959),

essentially by progressively heating and cooling

igneous rock samples and pottery fragments to

retrace the acquisition of their thermoremanence.

The Thelliers experimented with partial ther-

moremanence (pTRM), which is induced in a

sample that is cooled in a field through a limited

temperature range and in zero field for the

remaining temperature intervals. pTRM is iden-

tified with grains or domains that have blocking

temperatures within the range of the field expo-

sure. The experiments established the principle

of additivity of pTRMs: the pTRMs acquired over

all of the separate temperature ranges add to the

total TRM observed if the sample is cooled in

the field over the whole temperature range.

The pTRMs acquired in a low field (the Earth’s

field is low in this context) are independent and

each of them is lost by heating through the tem-

perature range over which it was acquired. The

432 ROCK MAGNETISM AND PALEOMAGNETISM

//FS2/CUP/3-PAGINATION/SDE/2-PROOFS/3B2/9780521873628C25.3D

–

433

– [417–437] 13.3.2008 10:45AM

Thellier method compares the pTRMs induced in

a sample by a known laboratory field with the

components of the natural remanence that are

lost by heating through the same temperature

ranges. It provides a consistency check by giving

estimates of the original field strength from the

remanence induced in each of several temper-

ature ranges. The high blocking temperature

components are generally the most stable and

so give the most reliable results.

Theideaisthat,ifasamplehasasimplether-

moremanence and is chemically unaffected by

heat, then the Thellier method gives an estimate

of the strength of the original field in which it

cooled; the error arising from the dependence of

blocking temperature on cooling rate is assumed

to be small (see Problem 25.2). However, there are

some doubts and difficulties. Chemical changes

during laboratory heating can usually be recog-

nized, but there is a possibility that the natural

remanence of an apparently unaltered igneous

rock is not a simple thermoremanence but, at

least partly, a chemical remanence, induced by

the field during development or modification

of the magnetic minerals. In these cases, the

pTRM observations generally indicate that some-

thing is wrong by disagreement between the

field estimates from different pTRM components.

Unfortunately rather few rocks give ideal results;

ancillary tests are applied to determine whether

field estimates from some components are never-

theless acceptable.

Several modifications of the Thellier method

are now preferred, in most cases designed to reduce

the labour-intensive procedure of repeatedly heat-

ing, cooling and re-measuring samples. The use of

field-free space is now standard, although not used

in the original work. Shaw (1974) supplemented

thermal measurements with alternating field

measurements. Anhysteretic remanence (ARM) is,

in some interesting ways, analogous to thermo-

remanence. ARM is induced by a small, steady

field applied while a superimposed alternating

field is reduced to zero from a high value, a process

that would demagnetize a sample in the absence of

the steady field. Partial ARM (pARM) is observed

when the steady field is applied only for a limited

range of the diminishing alternating field. pARMs

are additive in the same way as pTRMs and the

‘spectrum’ of ARM demagnetization is often very

similar to that of TRM. Shaw (1974) showed that

comparison of the alternating field demagnetiza-

tion curves of a sample before and after heating

allows identification of parts of the curve that are

unaffected by heat and are assumed to give reliable

paleointensity estimates.

An assumption intrinsic to the Thellier

method is that the different pTRM components

are independent. Dunlop and

Ozdemir (1997)

drew attention to the fact that the additivity of

pTRMs is essentially a single domain pheno-

menon, with independent fine grains having indi-

vidual blocking temperatures. In large grains the

individual domains may be blocked at particular

temperatures, but they are not independent, so

that true multidomain TRM is not additive. The

most stable remanence, which is of interest to

paleomagnetism, is of the pseudo-single domain

type (Section 25.2) and the assumption is that

this follows Thellier’s additivity principle well

enough to allow paleointensity measurements

using pTRM measurements.

Relative field intensities can be inferred from

sediment cores, as in Fig. 25.7, either by assum-

ing that the sediment is uniform over the depth

range of interest or by using susceptibility as a

measure of the abundances of magnetic miner-

als. Absolute intensity measurements by the

Thellier or Shaw methods necessarily rely on

igneous rocks with simple histories and suitably

small magnetic grains. These are increasingly

difficult to find as one looks further back in time.

It is the oldest rocks that are of greatest inter-

est, as indicators of the very early field. Available

data for the age range 2.0 to 3.5 billion years

(Hale, 1987; Halls et al., 2004; Macouin et al.,

2004; McArdle et al., 2004) suggest that the field

was then systematically weaker than in more

recent times, as might be expected from the dis-

cussion of dynamo power in Section 22.7. It is

more or less impossible to be sure of paleointen-

sity results from such old rocks and the field

varies continuously anyway, so an enormous

amount of data would be needed for a secure

conclusion. But there is one important conclu-

sion that is secure: there has been a field com-

parable to the present one continuously for at

least 3.5 billion years.

25.5 PALEOINTENSITY 433

//FS2/CUP/3-PAGINATION/SDE/2-PROOFS/3B2/9780521873628C25.3D

–

434

– [417–437] 13.3.2008 10:45AM

25.6 Polar wander and continental

drift

Direct observations of the Earth’s rotational axis

by conventional astronomical methods, and now

more accurately by satellite observations and very

long baseline interferometry (VLBI), show that

there is a drift of the North Pole towards 79 8W

at about 11 cm/year. Strictly, it is the Earth’s fea-

tures that are moving relative to the rotational

axis, which is unaffected by this process. The

motion is superimposed on the 14-month and

12-month wobbles that have a total amplitude of

about 5 m (Section 7.3). The polar drift is attri-

buted to an axial readjustment by the mass redis-

tribution of post-glacial rebound (Section 9.5), a

consequence of asymmetrical glaciation of polar

regions during the last ice age. It is a transient

phenomenon when viewed on the time scale of

the tectonic processes that cause polar wander

and continental drift, and causes a total move-

ment of the pole of no more than a few kilo-

metres. However, the rebound drift obscures

from direct observation by modern geodetic

methods the longer-term polar migration that is

studied by paleomagnetism. Satellite and VLBI

methods are used to measure tectonic move-

ments of the plates relative to one another, but

cannot distinguish the absolute tectonic motion

from the rebound effect.

On the other hand, paleomagnetism measures

continental movements, relative to the pole, over

hundreds and even thousands of millions of

years. The paleolatitude of a rock at the time of

its formation is given by the dip angle of its mag-

netization, using Eq. (24.11), and its orientation is

indicated by the direction of the horizontal com-

ponent of magnetization. Thus, the angular dis-

tance and direction to the pole are determined

and its position, relative to the sampled rock, can

be plotted on a globe, or on a projection of one.

Sufficient observations are made to average out

the secular variation and, assuming validity of the

axial dipole principle (while acknowledging the

small far-sided effect discussed in Section 25.3),

the paleomagnetic pole determined in this way is

also the geographic pole. Measurements on a ser-

ies of samples of different ages from the same

land mass give a series of pole positions that

mark a path, called an apparent polar wander

path. Thus, a primary observation of paleomag-

netism is polar wander, which can be observed

unambiguously for any land mass that remains

coherent. A typical rate of polar wander is 0.3 8

per million years and, for rocks younger than

about 20 million years, this cannot be separated

satisfactorily from the scatter due to secular var-

iation (Fig. 25.4). Deviations from the present pole

are apparent at about 30 million years and

become clearer with increasing age.

Greatest interest in polar wander curves

arises from comparisons of pole paths for differ-

ent continents, as in Fig. 25.10. Since, by the axial

dipole principle, the averaged pole was in the

same position at any particular time for all con-

tinents, the differences between pole paths indi-

cate relative drift of the continents. But, whereas

polar wander, relative to a particular land mass,

is completely specified by paleomagnetic obser-

vations, continental drift is subject to an ambi-

guity in longitude. Given the position of a pole

for a continent, we can place the continent on a

globe in latitude and orientation, but its longi-

tude is arbitrary. Thus, the longitude difference

between two continents cannot be deduced

solely from paleomagnetic observations. For

continental movements over the last 150 million

years we have vital additional information

in the magnetic stripes on the ocean floors

(Section 12.3) and the relative movements of

continents are fully resolved. For earlier peri-

ods, from which there is no surviving ocean

floor, the longitude ambiguit y remains a formal

problem, and less precise methods are used to

resolve it. The record is less secure for the

Precambrian period, more than 500 million

years ago, but continental reconstructions back

to 2.5 billion years have been presented (Pesonen

et al., 2003).

The first two pole paths to be compared were

for Europe and North America (Fig. 25.10). As was

immediately recognized, the simplest explanation

for their divergence is that the two continents

were once adjacent and drifted apart by the open-

ing up of the Atlantic Ocean basin. This is an

interesting example of the longitude ambiguity

problem mentioned above. The curves are

434 ROCK MAGNETISM AND PALEOMAGNETISM

//FS2/CUP/3-PAGINATION/SDE/2-PROOFS/3B2/9780521873628C25.3D

–

435

– [417–437] 13.3.2008 10:45AM

explained by a relative longitudinal drift of the

two continental blocks, but paleomagnetism can-

not distinguish this from rotation. However, rota-

tion sufficient to explain the separated pole paths

would give overlap of the continents and must be

discounted. In any case we have marine magnetic

anomalies to demonstrate that the opening of the

Atlantic basin is the correct interpretation.

The displaced pole paths for Europe and North

America revived the idea of continental drift. It

was slow to gain general acceptance, but the sub-

sequent demonstration of the greater movement

of Australia relative to the northern continents

removed much of the lingering doubt. The idea

that the continents bordering the Atlantic were

once juxtaposed and rifted apart has a long history.

Similarities of continental shapes and the rocks on

corresponding margins provided the early drifters

with the evidence they used to argue for conti-

nental mobility. The early discussions focussed

mainly on the work between about 1910 and

1930 of Alfred Wegener, a German meteorologist.

Fig.25.11isamorerecentfitofthecontinents

bordering the Atlantic, joined in the manner that

Wegener favoured. Another important pioneer

was the South African geologist A. L. DuToit,

whose 1930s reconstruction of Gondwanaland,

the composite southern continent comprising

FI G U R E 25.10 Pole paths for Europe (open circles) and North America (solid circles) for a 250 million year period

from the Jurassic to the Ordovician. These pole paths can be made to coincide by closing the Atlantic Ocean and

joining the continents as in Fig. 25.11. Figure redrawn from Van der Voo (1990).

25.6 POLAR WANDER AND CONTINENTAL DRIFT 435

//FS2/CUP/3-PAGINATION/SDE/2-PROOFS/3B2/9780521873628C25.3D

–

436

– [417–437] 13.3.2008 10:45AM

South America, Africa, India and Australia in a

cluster around Antarctica, has been fully vindi-

cated by paleomagnetism (Fig. 25.12).

In early paleomagnetic work, when the sub-

ject was still contentious, it was considered

important to demonstrate agreement between

magnetically measured paleo-latitudes and

climate. Glaciation is a particularly effective cli-

matic indicator. Although the global climate is

variable, as ice ages demonstrate, it is consistent

at any one time. Thus, the ice age glaciations of

North America and Eurasia, as well as the south-

ern Andes and South Island, New Zealand, were

synchronous. They do not indicate very rapid

polar wander but simultaneous cooling. Heavy

glaciation occurred in the late Carboniferous

and Permian periods, and Holmes (1965) showed

that the evidence for this in the southern conti-

nents is consistent with their clustering about

Antarctica at that time (Fig. 25.13).

Plate tectonics (Chapter 12) is now central

to global Earth science, with a wide range of

observations to elucidate the details. But the

starting point was the paleomagnetic demon-

stration of continental drift and everything

else followed. The apparent relegation of the

FI G U R E 25.11 A fit of the 500

fathom (914 m) depth contour for

continents around the Atlantic

Ocean. Shaded areas represent

overlap and black areas are gaps.

Redrawn after Bullard et al. (1965).

436 ROCK MAGNETISM AND PALEOMAGNETISM