91É«ÇéƬ

Mountains grow where continents collide. The slow but steady shifting
of continental plates, bringing earthquakes, volcanoes and mountains where
the plates meet, has shaped the surface of the Earth. This theory-plate
tectonics-has also shaped geological thinking for the past 30 years. But
researchers are beginning to realise that the plate tectonic story is not
as simple as it once seemed, especially for the continents.

In the theory of plate tectonics, the surface of the Earth is made of
plates of lithosphere-crust and upper mantle-that continually shift their
positions. The plates are rigid, so places within a plate move as one; the
rocks there do not have to stretch or crumple. The movement between two
plates is apparent at their edges; the world’s volcanoes and earthquakes
cluster along the plate boundaries. The movements are slow, but measurable,
typically a few centimetres per year. Deep underground, rocks flow, and
the plate movements are steady. But we have no way of measuring this widespread,
deep flow. Instead, researchers interested in the deformation that results
from plate movements have concentrated on the upper part of the Earth’s
crust. Within 10 or 20 kilometres of the surface, rocks are cold and tend
to fracture rather than flow. We feel the larger fracturing events as earthquakes,
when the stress that builds up as the plates steadily move becomes too great
for the rocks, and they shatter. The greater the build-up, the more rock
shatters when the quake comes. Plates move at centimetres per year, and
big quakes can involve slip of a few metres every few hundred years. Measurements
of seismic activity, large and small, give a good idea of the strain that
the plates exert on the Earth’s crust. But researchers now realise that
the quakes are not telling the whole story.

In the late 1980s, James Jackson and Dan McKenzie, two earth scientists
at the University of Cambridge, added up the rate of plate movements in
various directions around the world and compared them with the sum of the
slip in earthquakes in the same places over comparable periods of time.
According to plate tectonics, earthquakes are the expression of the slip
between the plates, so the amount and rates of movement between two plates
should match the sum of the slip on all the quakes at the plate boundary.
But Jackson and McKenzie showed that they did not.

They found that in some places, the strain resulting from earthquakes
was less than the movement of the plates over the same time. The mismatch
between the two was greatest on the continents; the researchers concluded
that something else as well as earthquakes must be going on in the upper
crust.

The rocks now forming mountain chains are evidence for the importance
of other types of deformation in their past; rocks have flowed, spread and
folded. This slow, remorseless type of movement, called creep, is different
to that caused by earthquakes, which happens in abrupt speedy pulses of
slip on faults. Most creep comes from deformation deeper in the crust, where
rocks are hotter and under more pressure. But could creep be important in
the upper crust too, and add to the slip from quakes to match the plate
movements? And if it is, how does it happen where the rocks are cold and
should fracture rther than flow?

Old faults know best

One way to tackle this question would be to look at places where continents
are deforming now. In California, for example, creep is part of the earthquake
story; segments of faults that are creeping do not give big quakes. But
the rocks themselves are out of reach, even a few kilometres below ground.
Researchers in the US have drilled a core through one of the faults in the
San Andreas fault system, but they do not claim that this gives a representative
picture. We tried a different approach. We decided to look more closely
at an old fault system, one whose large-scale geological appearance already
provided clues to how far and how quickly the faults had moved. We could
compare these estimates with the detailed record of deformation processes
preserved in the structure of the rocks and their individual mineral grains.
And because rivers erode slices through mountain chains, we can see a reasonable
proportion of the rock, and have enough information to reconstruct the evolution
of the area. The first stage is to find out how long the deformation took.

The speedometer for plate movements on Earth comes from the ocean floor.
At mid-ocean ridges, volcanoes steadily erupt basalt; as the lava solidifies,
this new ocean crust slides away from the ridges as if on a conveyor belt.
But as they crystallise, iron-bearing minerals in the basalt become slightly
magnetised parallel to the direction of the Earth’s field. This direction
can reverse-at certain times in the past the North Pole was where the South
Pole is now. Deep sea drilling programmes have calibrated the patterns of
normal and reverse magnetism, by linking them to the sediments that settled
on top of the lava: the sediments contained fossils that dated the igneous
rocks beneath, and, in turn, the reversals. Magnetic minerals in sedimentary
rocks formed on land added to the sea floor pattern.

Lava with normal and reverse magnetism forms a symmetrical pattern of
stripes on either side of the mid-ocean ridge where it erupted. Once geologists
know the history of reversals, they can say when each stripe formed. The
lava formed at the ridge, so the shape of the stripe and the older ocean
crust on each side of it tell them the shape of the ocean at that time.
So we can re-run plate movements, pulling back the sheets of ocean crust
to the ridge and bringing the continents back with them.

But this method can only produce rates of plate movements averaged over
several million years. Satellite technology is now adding some detail to
the picture of infinitesimal movements. Through the Global Positioning System
(GPS), a navigational system based on receiving signals from satellites,
geophysicists have worked out how fast the plates are moving-now, over the
past decade, or over the past century. They can detect variations in movement
even in relatively small areas. Part of Greece around the Aegean Sea shows
this very clearly, thanks to the lucky survival of marker points from a
surveying network set up a century ago. An inter-university team of British
and Greek geophysicists resurveyed the surviving parts of this network,
using GPS (see Nature, vol 350, p 124). Some of the markers had hardly moved,
while others were a metre or so away from their earlier position. These
separations delineated the plate movements during this time, and found that
one area, about 120 kilometres from north to south, was roughly a metre
longer than it had been a century ago. But the known earthquakes during
this period accounted for only about 70 centimetres.

So what else is going on? To find out we need to combine geological
experiments with some theory from materials science, and compare the results
with what we see in naturally deformed rocks. This reveals close links between
the deformation of rocks and other materials with related crystal structures,
such as metals and ceramics . More importantly, rates of deformation estimated
from the microscopic structure of the rocks-evidence of creep-fill in some
of the gaps in the plate tectonic story.

The rocks we studied were part of the Chartreuse Massif, near Grenoble
in southeast France. Best known for its monastic order who live in seclusion
and produce the renowned yellow and green liqueurs, it draws geologists
for other reasons, notably the spectacular cliffs of folded and faulted
limestones in deep gorges cut by rivers. Some of these layers of limestone
are almost pure calcium carbonate, and deform just like laboratory samples,
so the results of experiments should apply. This is important because intrinsic
impurities such as mud can greatly alter the mechanical properties of limestones.
Another point in its favour as a place to look for traces of earthquakes
was that the rocks had always been in the upper part of the Earth’s crust;
if they had been buried deeply, the limestone of the gorges would have changed
to marble.

The first stage was to find out when the mountains formed. The Chartreuse
hills are part of the outer ramparts of the Alps, known as the Subalpine
Chains. They and the rest of the Alpine mountains are part of a larger chain
stretching as far as the Himalayas. The Alps mark where there was once an
ocean called Tethys stretching almost around the world. As the plate movements
changed, Tethys started to close. Ocean crust slid beneath the continental
crust on the southern shore. The continents on each side began to move closer
together, and eventually collided, causing the rocks to fold and fault.
This is how the Alps were formed. High up, the odd fragment of ocean crust
remains, but otherwise Tethys has gone. Magnetic stripes elsewhere in what
was the ocean floor give some clues as to when the Alps formed, but they
do not help to pin down when the Chartreuse was made.

For this, we had to look closely at the rocks themselves. The faults
which built this part of the Alps are a special type called thrusts. They
are inclined very gently, and run from deep levels in the inner part of
the mountain belt, to shallow depths, even emerging at the surface close
to the its edges. Sometimes they cut through folds in the rocks; they also
fold the rocks above them as they move. From a cross section, compiled from
observations of the cliff sections and mapping the position and orientation
of different layers of rock, you can see that the folds and faults form
a pattern.

In the Chartreuse, the harder pure limestones are separated by thinner,
softer layers of lime muds. Thrust faults run along the softer layers, only
occasionally cutting upwards and across the thicker, stiffer beds. When
they do, the rocks fold as they move over the bends in the surface of the
fault. From a cross section, we could see which folds had moved first by
matching the folds above to bends on faults below. Sometimes two or more
thrust sheets-slices of rock above a particular thrust fault-moved together,
piggyback-style. And by matching places where a fault sliced through layers
of a particular age, above and below the fault, we could see how far the
fault had travelled. Unravelling the folds and faults in order in this way,
we found how far each fault had slipped. Although some of the faults may
have moved farther, we had a minimum estimate for how far the rocks had
moved, which would help us to understand how fast the mountains grew.

The next stage was to find out how long it all took. For this we turned
to sedimentary rocks that formed at the same time as the mountains were
growing. In the Alps these are pebbly sandstones and siltstones, and they
contain fossils that give precise ages for when they formed. In the Chartreuse,
some of the thrust faults cut across these sandstones, or fold them as the
thrust sheet is folded when the fault moves. So these faults must have formed
after the sandstones. This gives a starting date for the deformation. At
the end of the folding and faulting, the whole area was gently uplifted
and eroded. The erosion surface and some residual soils have been dated,
giving a time at which all the deformation had stopped.

In this way, we worked out that rocks of the Subalpine Chains had become
30 kilometres narrower during the Alpine deformation, until they reached
the present size (about 20 kilometres across). From the ages of the sedimentary
rocks that bracket the deformation, all this shortening took place between
12 and 6 million years ago. So the rate of shortening was at least 30 kilometres
in 6 million years, or 0.5 centimetres a year-even assuming continuous deformation,
and not including occasional faster spurts.

This is fast enough for earthquakes to have played a big part in the
formation of the Chartreuse mountains. In the San Andreas fault system of
California, the plates move at about 5 centimetres per year and there are
many small or moderate quakes each year. So it seemed as if the structure
of the Chartreuse, like California, could have come almost solely from earthquakes-except
for the evidence preserved in the rocks.

The limestones of the Chartreuse told a very clear story. Most of the
rock away from the major faults seemed untouched by the deformation. The
30 kilometres shortening had left fossils of coral with their original shapes
and internal structures, and fine layering that developed when the limestone
was laid down has survived intact. This is telling evidence against major
quakes; they and their aftershocks shatter rock over tens of kilometres.
Although we did see some shattered rock, it added up to only a small fraction
of the rock in the Chartreuse as a whole.

Even where the rocks were deformed, we found little evidence of extensive
shattering, or of the high stress that typifies quakes. The folds, which
seemed smooth and continuous on a large scale, proved on closer inspection
to be cut by many small faults. Some folds had started to form before the
faults, some after, but in both cases, the rocks bent by forming many small
faults. Imagine folding a telephone directory in half: where the rock was
made up of many thin layers, 20 or 50 centimetres thick, the folds developed
along the layers, just as each page slips past the next. Within the layers,
or where the bedding was thicker, folding was difficult, more like folding
a book in which the pages are thick card; it would be easier to tear a few
pages in half and slip them one on top of the other. In the rocks, faults
only metres long formed in the cores of the major folds, and slipped by
millimetres or centimetres. Some of the limestone dissolved as well, taking
up strain by pressure solution .

All the minor faults we found told the same story: the deformation was
brittle (fractures and cracks) but it was not the extensive shattering that
significant earthquakes bring. In fact, the minor faults were covered in
fibres of the mineral calcite-calcium carbonate-all grown in the same direction.
These fibres marked the direction of movement of the faults, and told us
that the slip had been very slow, compared to most earthquakes.

When we looked closely at the fibres, each one apparently a single crystal,
we found that they contained regular, closely spaced bands of tiny bubbles
a few hundredths of a millimetre across. In 1980, John Ramsay, professor
of geology at the Swiss Federal Institute of Technology in Zurich, recognised
this feature in other rocks; the bubbles represent planes along which the
rock had broken. As soon as the rock cracked, and the two sides of the fault
moved farther apart, the crack healed and a new layer of calcite grew on
the fresh surface. The cracking, and hence the slip of the fault, must happen
slowly enough for healing to take place. The rate of healing depends on
how fast calcium and carbonate ions diffuse along grain boundaries, or through
fluid that fills cracks and fissures in the rock. Diffusion through fluids
is much faster than along grain boundaries, but it is still too slow-in
an earthquake, rock moves about a million times as fast.

Upper crust creep

These minor faults could not have produced any sizeable earthquakes-they
are too small, the rock adjacent to them was in places pristine and they
formed and slipped too slowly. What they can do is explain how creep could
happen in the upper crust. Slip on faults a metre long and smaller can result
in smoothly varying, continuous strain on a larger scale. Slow steady cracking
and healing spreading across the upper crust could be the key to continental
deformation.

Not all the fracturing in the Chartreuse was quite this slow. Some fibres
were broken up into fragments held together by larger areas of clearer calcite
without the tell-tale bubbles. Instead, this calcite had a ring of crystals
at its edge, pointing beautiful perfect crystal faces into the centre, where
less than perfect crystals filled in the rest. These represented bigger
slip events. In some places, signs of shattering were more extensive, spreading
across volumes of rock of half a cubic metre or so. Even here, however,
the fragments stayed close together, and were cemented in place by calcite
crystals. The big thrusts were the most promising targets in our search
for quakes. Here, the rock was sometimes shattered for up to a couple of
metres around the fault zone, but in many places the damage was less extensive.
In general, the thrusts were layers of very fine-grained limestone, the
sort of rock that experiments have shown is very soft relative to undeformed
limestone-certainly too soft to allow stress to build up enough to give
a big quake.

We decided that despite some shattering-evidence for a sudden release
of energy in an earthquake-what quakes there were could not have been very
big. An earthquake happens when stress builds up enough to break rock apart;
the more energy stored up and released, the greater and more widespread
the damage. What earthquakes there were in the Chartreuse did not release
enough energy to shatter more than a few cubic metres of rock at a time.
It seems that the big quakes, if any, happened only as the thrusts started
to form, perhaps as they cut across the strong, thick limestone layers.
Once the slip had begun, the original rocks had been ground down so much
that, along the thrusts, they were softer than the rock in the surrounding
thrust sheets. As the rocks were pushed farther together the faults slipped
easily, and the rocks in the thrust sheets were carried on and upward, collectively
uplifting the Subalpine Chains to be eroded to the mountains we see today.

Once we had confirmed that earthquakes were only a part of the story
for this mountain chain, we looked again at the pattern of plate movements.
The processes we found in the Chartreuse, although slow, could easily keep
pace with the plate movements when combined with some small quakes, and
could have built the mountains in the 6 million years that we know it took.
But imaginary observers during those 6 million years would have missed most
of the story if they had relied solely on a few seismometers. Creep is an
important extra factor in seismically active areas, where it takes up the
strain along stretches of faults that do not spawn big earthquakes. What
is different about our results is that there appear to have been few, if
any, big quakes. We found no traces of cataclasis on the scale envisaged
for magnitude 7 or 8 earthquakes-California’s Big Ones. The lesson from
the Subalpine Chains is that earthquakes need not be an important part of
deformation on the continents; tiny fractures and faults that seem insignificant
at first sight can have an equally large effect.

Satellite measurements, seismometers and other remote sensing devices
are refining our picture of the details of plate movements and interactions
today, but there are processes that they cannot reveal. These processes
are not necessarily obscure. Most of them have happened before in the 4
billion year history of the planet; the way forward is to go and look for
those processes in the rocks. The full story, or as much of it as we can
hope to have, will come if we take the present as the key to the past and
use today’s geophysical data in conjunction with the history held within
the rocks.

Rob Butler is a geologist who works on the structure of mountain belts
at the Univeristy of Leeds.

* * *

Rock and rolling mills – the signs in the stones

Metallurgists study how to process a metal or alloy to influence its
crystal structure and physical properties. Geologists do the opposite: they
take a deformed rock, and find out how it formed. In the 1970s, geologists
realised that ideas from materials science could tell them whether rocks
had come to their final shapes at high or low temperatures, quickly or slowly,
wet or dry.

A simple example comes from two methods of steel production-hot working
and cold working. Each leaves behind distinct imperfections in the crystals
of the metal. Cold working consists of rolling steel at relatively low temperatures.
Individual crystals distort to take up the strain as the sheet becomes thinner.
Planes of atoms in the crystals slide past each other, breaking and reforming
bonds between individual atoms. No atom moves far from its mean position,
but the pattern of bonds alters to form linear breaks in the otherwise regular
structure, called dislocations.

As the rolling continues to stress the crystal lattice, bonds between
atoms break and reform and the dislocations move, on different planes within
the lattice. When two or more dislocations that are moving through the lattice
intersect, they become tangled up, and neither can move farther. Sometimes
dislocations moving on the same plane pile up; otherwise the tangles localise
where two lattice planes intersect. As a result, the steel becomes harder
as it is worked.

Hot working has a different effect. At higher temperatures dislocations
form and move through the lattice, but they do not tangle up. Because the
steel is warmer, atoms in the lattice have more energy, and bonds can move
in different ways to relieve the local stresses around a dislocation. In
particular, bonds can break and reform all around the defects, so that dislocations
are no longer restricted to their own planes. They can continue to move
without tangling, and the steel can reach much higher strains without hardening.

Metals worked in these ways look very different under a transmission
electron microscope (TEM): cold worked steel has dense areas of tangled
dislocations throughout the crystals. Hot working leaves many areas of the
crystal with few dislocations; what defects there are form the edges of
these subgrains. These areas have not been protected in some way from distortion.
They have been distorted, and have subsequently recovered; this is why hot
working can achieve higher strain.

Geologists find these two types of deformation interesting for two reasons.
First, they can tell them apart; secondly, they represent relatively cool
and hot deformation. Likewise, other deformation mechanisms produce distinguishing
features that might indicate high stress, or fluid pressure or even the
strain rate. These microstructures revealed the processes that deformed
rocks such as mylonites.

Mylonites are tough, banded, fine-grained rocks that geologists had
associated with fault zones for more than a hundred years. Charles Lapworth,
of the University of Birmingham, realised in the late 19th century that
the peculiar textures of mylonites had something to do with fault movements.
He noticed that the texture of a rock changed gradually the nearer it was
to a fault. No matter what sort of rock he looked at, they all developed
the banding and became fine grained near movement zones. Lapworth thought
that the very small grains in mylonites were the result of grinding by the
movement on the fault-a fair guess, but an idea that Stan White, a geologist
at Imperial College London was able to disprove a hundred years later, thanks
to the technology of electron microscopy.

Looking at rocks from the same area that had fascinated Lapworth, the
northwest Highlands of Scotland, White was struck by how much the microstructures
of these rocks resembled textures of hot-rolled steel, especially in TEM.
Mylonites were clearly products of a hot working process, at relatively
high temperatures. Lower temperature deformation could crush and grind down
rocks into fragments, but without the same dislocation structures.

So the search for different mechanisms of deformation, and the physical
conditions they represented, began. Geologists realised they could produce
some textures that matched real rocks in a matter of a few months. But how
could experiments hope to reproduce millions of years of infinitesimally
slow geological deformation? One way was to concentrate on the speedier
aspects of deformation-fracturing and crushing. Another was to chose softer
rocks such as limestone, or to run experiments at higher temperatures.

Experimenters try to use relatively simple rocks such as pure quartzites,
limestones, and granite with uniform grain sizes. They mount small cylinders
a few centimetres high in presses and surround them with fluid under pressure
to mimic the pressure underground from surrounding rocks. The results have
helped quantify the processes behind some common features of naturally deformed
rocks.

One important mechanism, albeit very slow, is diffusional mass transfer,
in which minerals dissolve at some places and precipitate elsewhere, changing
the overall shape of the rock. This process makes ice skating possible:
ice, like minerals such as calcite, is more soluble when compressed. Beneath
the blade of a skate, ice melts to make a slippery film of water, on which
you glide. Once the pressure drops, the water freezes back into ice. In
rocks, grains tend to dissolve fastest in the direction of the greatest
compression. The effect is especially important in limestone and dolomite;
it often happens at different rates through the rock, because of impurites
and variations in crystal structure.

Cataclasis (fracturing) is also important, especially in the top 10
kilometres or so of the Earth’s crust. Fractured rocks are often also permeable;
this can speed up further fracture, by bringing chemicals such as water
that weaken the rock to the vulnerable tips of cracks. It also contributes
to faster weathering of fault zones, which often form valleys.