Download by William J. Crornie Rapidly developing technologies are

by William J. Crornie
Rapidly developing technologies
are adding to the ways
earth scientists can examine
their enigmatic subject.
28 MOSAIC
Geologists are getting deeper into
their work. Once they were
confined largely to the earth's
surface; now new techniques are opening windows that let them see all the
way to the planet's center. The new
views are startling, if somewhat blurry.
They show internal details that are reflected in the formation and drifting of
continents, in the making of mountains
and islands, and in volcanism and
earthquakes. These details eliminate
many cherished explanations of such
processes, and they set narrower limits
to explanations that are possible.
As recently as 1960 most geologists
believed that the continents had remained fixed throughout the 4.5-billionyear history of the earth. They thought
that mountain building, earthquakes,
and volcanism involved vertical movements, up and down adjustments of an
elastic interior to changes in mass at the
surface. Geologists then described the
planet as having an outer crust of solid
rock varying in thickness from 10 kilometers under the oceans to 70 kilometers under the continents. They
compared this crust to the shell of a
hard-boiled egg u n d e r which lies a
white (the mantle) about 2,800 kilometers thick. The yolk (core) extends
from 2,900 to 6,400 kilometers and consists of a 1,100-kilometer-thick, semifluid envelope enclosing a solid, iron™
dominated inner core.
By 1970, it was difficult to find a
geologist who did not believe that
oceans open and close, and that continents split apart, drift, rotate, and collide with each other. Virtually every
earth scientist championed the scenario
demanded by the theory of plate tectonics, which holds that the earth's
shell, rather than being seamless, is
broken into 20 or so plates, the largest of
which consist of continents attached to
huge sections of ocean floor. Oceans
open and close and continents drift because the plates move with respect to
each other.
Heat escaping from the mantle causes
molten material, or magma, to rise and
rift the thin crust that underlies the
oceans. This creates submarine valleys
and ridges made of the youngest rocks
on earth. The peaks of the ridges sometimes thrust through the surface as volcanic islands. The ridges also mark the
inner boundaries of plates that are moving away from each other.
The outer, leading edges of spreading
plates collide with each other. When
they converge, the edge of one plate
may override the other, forcing the edge
of one plate down into the mantle and
producing mountains, volcanism, and
earthquakes. Converging plates also
slide by each other. California is split by
a plate boundary, visible at the surface
as a series of faults running from the
Gulf of California to San Francisco.
Stresses along the faults relieve themselves as earthquakes.
The National Academy of Sciences'
Committee on Opportunities for Research in the Geological Sciences has
called the theory of plate tectonics "a
compellingly attractive hypothesis, simple, elegant, and potentially able to explain a wide range of diverse observations." The theory explains continental
drift, the appearance and disappearance
of oceans, and the observation that no
rocks older than 200 million years have
ever been dredged from the deep sea.
(Two hundred million years is the maximum time it takes for a rock born at a
mid-ocean ridge to be destroyed or
changed by plate convergence.)
Plate tectonics theory also accounts
for the positions and surface features of
continents dating back at least 600 million years. Beyond this time, the history
of the earth becomes, in the words of
Charles L. Drake, a geologist at Dartmouth College, "more theological and
less geological." The major reasons involve the ages of continents.
Radioactive dating shows that continental rocks are as old as 3.8 billion
years. In contrast to the sea floor's
straightforward creation, spreading,
and destruction, the continents have
been repeatedly compressed, extended,
split apart, rotated, drifted, and subjected to wave after wave of collisions
with other continents, islands, a n d
slabs of ocean floor. To unravel this chaotic history, geologists need to extend
the plate-tectonics revolution from sea
to shore, from 600 million to 4 billion
years, and from the crust to the core of
the planet. (See ''Tectonics in an Archean Earth," in this Mosaic.)
A deep hole
Deep drilling provides the most direct
view of the earth's third and fourth dimensions, depth and time. It provides
only a shallow view, although Soviet scientists have shown that it can be a highly revealing one. The Soviet Union's
exploratory drilling program includes
the deepest hole drilled to date. Spudded in the Kola Peninsula east of
Finland, the hole reportedly goes down
13 kilometers, 85 percent of the way to
its targeted depth. Although the United
States leads in deep-ocean and commercial drilling, its deepest hole on land,
drilled by the United States oil industry,
bottomed at 9.6 kilometers.
The Kola bore hole is one of a series
being drilled to obtain both practical information and basic knowledge of the
earth. Since its drill bit first chewed into
rock in 1970, the Kola effort has been
producing data that cast doubt on traditional theories of the structure of the
continental crust, heat flow, and mineral
formation.
Textbooks describe the continental
crust as being composed of silicon- and
aluminum-rich granitic rocks. These
overlie denser basaltic rocks, which are
lower in silicon and higher in magne™
sium. Geologists draw this picture on
the basis of a sharp increase in the velocity of earthquake waves as the waves
travel downward into the crust.
Soviet researchers intended to penetrate the so-called Conrad discontinuity
between the two rock types and bring
up samples of each. Minister of Geology
Yevgeny A. Kozlovsky has r e p o r t e d ,
however, that "with increasing depth In
the Kola hole, the expected increase In
rock densities was not recorded. Neither
was any increase in the speed of propagation of seismic waves nor any other
changes in the physical properties of the
rocks detected. Thus the traditional idea
that geophysical data obtained from the
surface can be directly correlated with
MOSAIC 29
Eurasian
Crustal plates jostle each other on earth's surface. Spikes point in the direction of subduction where they meet.
geological materials in the deep crust
must be reexamined/
"It's a matter of interpretation/' says
Jack Oliver, a seismologist at Cornell
University. "All rocks are compressible,
and at those depths pressure may be responsible for giving certain kinds of granitic rocks seismic characteristics not
unlike those of basaltic rocks." And, he
cautions, "you must always take mistakes in measurement into account."
"What really surprised us about the
Kola hole results," Oliver says, "is that
they found open fractures with fluids
flowing through them at depths down
to at least 11.5 kilometers. Many scientists expected that pressure would close
all open space below three kilometers."
Fluids in the open spaces may carry valuable minerals, which precipitate out as
the liquids rise and cool. Soviet drillers
encountered what they say are commercial quantities of copper, cobalt, zinc,
lead, and iron between depths of 1.7
and 6.5 kilometers.
The same fluids bring up unanticipated amounts of heat. ' T h e temperature gradient measured down to
three kilometers corresponded to that
expected, i.e. one degree Celsius inCromie, executive director of the Council for
the Advancement of Science Writing, writes
frequently on the earth sciences.
30 MOSAIC
crease per 100 meters depth," Kozlovky
reported. "Below this, the gradient increased to 2.5 degrees per 100 meters,
and at ten kilometers the temperature
was 180 degrees Celsius instead of the
expected 100 degrees."
Evidently enough fluid flows to transfer heat by convection, a much more
efficient process than conduction
through solid rock. "We once believed
that heat was transferred through the
ocean crust entirely by conduction," the
Cornell seismologist says. "Now we
know that the major transfer occurs
through convection at mid-ocean
ridges. Confirming Kola hole results
may mean that the same process operates in the continental crust/
Mountain building
Scientists in the United States hope to
get their own deep continental drilling
program started by 1986. The White
House Office of Science and Technology
Policy has recommended that funds for
site surveys and drilling be appropriated in fiscal years 1985 and 1986.
The Continental Scientific Drilling
Committee of the National Academy of
Sciences has assigned the highest priority to a ten-kilometer hole in the
southern Appalachian Mountains to
test a new concept of mountain building. Geologists suspect that the Appala-
chians and many other mountain belts
were thrown u p w h e n plate convergence caused huge slabs of rock to be
thrust up and over a continent. Some of
the overthrust sheets evidently traveled
hundreds of kilometers, bulldozing seafloor sediments and continental margins into coastal mountains. (See "Old
Rock on Young Rock" and "Seismic-reflection profiling" by Henry Simmons,
Mosaic, Volume 14, Number 2.)
The idea that lateral thrusting may
have happened at m a n y places and
times in the earth's history has existed
since the last century. Evidence that it
actually occurred, however, was seen
only recently through another window
into the planet: seismic-reflection profiling. This technique involves beaming
long trains of sound waves, or vibrations, of varying frequencies into the
earth. The frequencies vary from 8 to 40
cycles per second. The wave trains reflect vertically from faults and rock
boundaries and are picked u p by an
antenna-like array of geophones implanted at the surface.
In 1974, a group of researchers from
Cornell and other universities organized
the Consortium for Continental Reflection Profiling, or Cocorp, to apply this
technique to study the deep basements
of continents. To date the consortium
has completed more than 5,500 kilo-
meters of seismic surveys in the United
States. Maps of reflection horizons
across the Appalachians in the Carolinas and New England yield evidence
of multiple, shallow, eastward-dipping
faults along which sheets of rock were
thrust westward by the closing of the
Atlantic and the convergence of continents between 250 million and 500 million years ago.
"Because the Appalachians are considered a representative mountain belt,
it is likely that thin-skinned overthrusting took place in many other locations/'
Jack Oliver states. These locations include the Ouachita Mountains in
Arkansas, the Rocky Mountains, the
Alps, the Himalayas, and mountains in
northwest Africa that appear to be mirror images of the Appalachians.
To be positively identified, a reflection
boundary must be traced downward
from the surface or sampled by drilling.
"The process seems so general that it is
very important to test the interpretations based on the seismic data with a
deep drill hole," says Oliver.
Mosaic continents
Overthrusting and buildup of continental margins also occurs on a lessthan-continental scale. The inexorable,
conveyor-like movement of ocean floors
toward destruction u n d e r continents
carries pieces of many sizes, including
chains of islands, sea m o u n t s , submerged plateaus, and miniature continents.
Geologists envision the western part
of North America as having been
formed by such a tectonic conveyor belt.
"Large-scale, overlapping thrust sheets
are well documented from Alaska to
southern California," says Allan Cox, a
geophysicist at Stanford University.
"The amount of differential movement
along some of these faults must certainly exceed several h u n d r e d kilometers." Cox counts 150 slabs and
blocks of "varying size that are as different from each other as they are from
land farther east/
The Basin and Range, a region that
lies between Utah's Wasatch Range and
the Sierra Nevada, now appears to be
undergoing extension along the same
type of low-angle faults that probably
welded it to the continent by compression hundreds of millions of years ago.
At that time, a large sheet of rock evidently slipped or was pulled back down
an inclined fault without breaking,
something thought previously to be impossible. Surveyors for Cocorp have
traced such a fault 70 kilometers into the
crust, slanting down to a depth of 25
kilometers. Such a situation provides an
exceptionally clear look through the
seismic window; geologists do not have
to guess what kind of boundary reflects
the seismic energy because boundary
rock is exposed at the surface.
The Cocorp researchers believe that
the continental crust has stretched
along these faults to double its original
area, adding as much as 300 kilometers
to the distance between Salt Lake City
and Reno. This stretching is accompanied and may be caused by the rise of
magma from below. Surface swelling
and the behavior of seismic waves indicate the presence of chambers of molten
material under New Mexico near Socorro, under Death Valley, and under Long
Valley in the Sierra Nevada of eastern
California. Geologists blame movement
of magma for recent earthquakes in the
Long Valley area. The magma pushed
the valley floor upward 45 centimeters
in five years and prompted the U.S.
Geological Survey to issue a volcanohazard alert in 1982.
To open the seismic window wider,
reflection profiling is used in combination with refraction, a technique that
measures the speed of shock waves that
travel horizontally through rocks rather
than being reflected from boundaries
between them. These waves eventually
are bent or refracted back to surface
recorders. Their velocities depend on
the type of rock traversed, and the velocities can help identify different rock
types.
Refracted waves travel more slowly in
the mantle beneath the Basin and Range
region than they do under the more stable crust to the east. Some researchers
interpret this to mean that the western
part of the crust is stretching along a
flat, semiliquid layer that marks the
crust-mantle b o u n d a r y at a d e p t h of
about 25 kilometers in this region. They
picture blobs and chimneys of magma
spreading out along this boundary after
rising from a layer of soft or lowviscosity material underlying all the
plates at a depth between 50 and 100
kilometers. Heat may cause the Mohorovicic seismic discontinuity, or Moho,
the crust-mantle boundary, to flatten
out in this area like hot asphalt.
K. Douglas Nelson, a geologist at Cornell University, speculates that the
Moho may also represent the maximum
height to which basaltic magma from
under the plates normally rises. The
less-dense crustal material above acts as
a lid, causing the rising magma to
spread out at the base of the crust.
Where the crust is extending, however,
magma can continue to work its way up
from the Moho to the surface. Still to be
answered is the question of whether rising magma is the cause or the result of
crustal extension.
Hot spots and undulations
Shallow magma bodies, like the ones
under Death Valley and central New
Mexico, move with the continents.
Plumes, like the 300-kilometer-deep one
that generates the geysers of Yellowstone National Park, remain relatively
stationary as the plates drift over them,
thereby serving as references for measurement of the motions of the plates.
Earth scientists have mapped 42 hot
spots, most of them lying under the
oceans and Africa. Typically the hot
spots are crowned by volcanoes and sit
near broad domes or swells in the crust.
The island of Hawaii, for example, with
its recently active volcanoes, sits on the
edge of such an uplift about 1,500 kilometers wide" and one kilometer high.
Many geologists believe that magma rising from the mantle in these areas initiates rifts that break apart continents and
ouen UD new oceans. (See "The Earth's
Foundations of Fire/' by Patrick Young,
Mosaic, Volume 11, Number 3.)
Hot-spot swells and other bumps and
depressions in the surface can be
mapped through a window opened by
earth satellites. Radar altimeters on orbiting spacecraft record small- and
large-scale variations in sea level. These
variations are reflections of gravity
anomalies, and they can be correlated
with changes in the elevation of the
ocean floor.
Uplifts in the vicinity of the Hawaiian
Islands and other hot spots add extra
mass to the surface and create positive
anomalies. Negative anomalies occur
where cold, dense material causes depressions.
The anomalies can be smaller or
larger than the l,500-to-3,000-kilometerwide hot-spot swells. William F. Haxby
and Jeffrey K. Weissel, geophysicists at
Lamont-Doherty Geological Observatory in New York, examined data from a
radar altimeter aboard the Seasat spacecraft and found smaller-scale anomalies
MOSAIC
31
beneath the Pacific and Indian oceans.
High and low anomalies alternate,
with a wavelength of 150 to 220 kilometers. Haxby and Weissel attribute
these to convective motion in the relatively soft upper layer of the mantle on
which the plates move. The negative
anomalies may be attributable to cold
material sinking from the bottom of the
plates. This material would be replaced
by hotter, less-dense material, which
would create the positive anomalies. Evidence that the upper part of the mantle
is soft enough to permit this kind of motion includes the fact that seismic waves,
the speed of which decreases with decreased density, slow d o w n in this
zone. Called the asthenosphere (from,
the Greek word asthenos, meaning without strength) this zone extends downward to about 250 kilometers.
The 150-to-220-kilometer-long anomalies exist in areas adjacent to ocean
ridges, or diverging plate boundaries.
Between Central America a n d the
Hawaiian Islands, they a p p e a r to be
elongated, or smeared out, in the direction of plate motion, or west-northwest
relative to the hot spot under the island
of Hawaii. Haxby and Weissel find such
anomalies in the southeastern Pacific
Ocean and the Indian Ocean, too. These
also are areas where y o u n g crust is
being made at active ocean rifts. According to Haxby, 'The best explanation that
satisfies ail of the observations is that the
gravity-anomaly undulations reflect
small-scale longitudinal convection produced in a thin, weak, low-viscosity
layer of the upper mantle by shear imparted by the moving p l a t e s / ' Convection is driven by temperature differences, explains Barry Parsons of the
Massachusetts Institute of Technology.
"The shear aligns convection with the
direction of plate motion," he says.
Not all earth scientists agree with this
explanation of the anomalies. What
does seem acceptable to everyone,
however, is that material from the crust
sinks where plates converge. The material descends, melts, and rises to the
surface again at mid-ocean ridges. On
the way it may get swept up in swellings
1,500 to 3,000 kilometers wide that are
associated w i t h hot spots, or in the
smaller-scale undulations postulated by
Haxby and Weissel.
Depressions are associated with thick
slabs of cold crust sinking into the mantle where plates converge. Highs,
however, do not coincide with hot mate-
32 MOSAIC
rial rising in the ocean ridges. Rather
they are centered over the equator, the
Atlantic-Africa region, and the central
Pacific. Lows are concentrated in a
broad band that Includes northeastern
Canada, the western Atlantic Ocean,
Brazil, Antarctica, the Indian Ocean,
and Siberia. This band includes 60 percent of the earth's total land area. One
theory is that continents slide from high
to low areas.
Don L. Anderson, a California Institute of Technology seismologist, proposes a model in which the thick
continents prevent heat from escaping
from the mantle. The heat buildup
causes crustal bulges. Eventually hot
spots associated with the upwarping
split the crust. Continents break apart,
and the pieces slide downhill to low
areas overlying sunken slabs of old
crust. The east coast of North America
may overlie the former Pacific Ocean
floor, which subducted under the westward-moving continent. In the future
the Yellowstone plume may split the
North American continent.
Tomography
Earth scientists obtain another view of
gravity anomalies through a new window that gives them a look at the earth
In three dimensions. The technique is
similar to computer-assisted tomography (a term based on the Greek word
tomos, meaning section.), wherein a computer assembles a series of X-ray images
having common points Into a three-dimensional picture. Earth tomography
replaces x rays with earthquake-generated waves that pass through the planet
or sweep its surface.
The longer the wavelength of surface
waves, the deeper they penetrate. Don
Anderson and his colleagues at Caltech's Seismological Laboratory use surface waves with periods of u p to 250
seconds to take tomographic slices
down to 400 kilometers. These waves
travel rapidly through negative anomalies, or anomalous lows, and more slowly through positive anomalies, or highs.
This difference is interpreted to mean
that the lows are cold areas of higherdensity rocks. Slow travel indicates hotter, lower-density material.
Sluggish travel occurs around the Yellowstone plume and around hot-spot
swells such as the one under East Africa, where up swelling material is splitting apart the continent and opening
the Red Sea. A l t h o u g h mid-ocean
ridges do not produce surface highs,
seismic waves travel slowly down to 200
between and 400 kilometers below these
features, indicating that hot material extends to those depths.
Surface waves from large earthquakes
also provide information about the direction of flow of convecting material in
the mantle. Seismic energy propagates
at different speeds in different directions, a phenomenon called anisotropy.
The anisotropy is attributed to the preferred orientation of crystals of olivine,
generally accepted as a principal rockforming mineral in the upper mantle.
Orientation indicates direction of flow,
Confusion range
House range
Sevier desert
Canyon range
and the seismic waves travel faster
through crystals aligned in this
direction.
Such data are interpreted to show that
material moves upward along the MidAtlantic Ridge, the East Pacific Rise, and
spreading ridges in the Indian Ocean
and the southern ocean around Antarctica. At 450 kilometers, flow direction
and propagation velocities suggest that
the flow is horizontal and the mantle is
cold and dense.
A more penetrating wiew
Body waves, which pass completely
through the earth, can be used to study
deeper parts of the mantle. Two Harvard University geophysicists, Adam
M. Dziewonski and John H. Woodhouse, combine data from surface and
body waves to construct composite images down to 2,900 kilometers, the mantle-core boundary. Because of the locations of seismometers, resolution with
this technique is limited to about 3,000
kilometers, and only large-scale anomalies can be mapped. "Nevertheless,"
Dziewonski says, "we have a tool with
which we are seeing things we did not
know existed a year ago."
"The most important conclusion," he
says, "is that below 400 kilometers,
maps of the mantle bear little relation to
surface features/' Indications of continents and ocean ridges begin to disappear at 200 to 250 kilometers. Various
ridges and sections of ridges can be
traced to different depths, and the data
suggest that ridges can be fed from the
sides rather than from directly below.
All evidence of ridges and the distinction between continents and ocean
basins, however, is gone at between 350
and 450 kilometers.
Below these depths a broad region of
high velocity underlies South America,
the South Atlantic, and West Africa,
suggesting the presence of cold, subducted rock. Seismic waves also travel
rapidly below the western Pacific, but
low speeds persist in the central and
eastern Pacific. Only hot spots, such as
the one associated with the northwestern Indian Ocean, retain their
character below 500 kilometers. ' T h e
hot spot associated with Hawaii may extend all the way to the core-mantle
boundary," Dziewonski notes. "In both
the upper and lower mantle, velocity
anomalies are significantly correlated
with the long-wavelength anomalies."
"The simplest assumption," Woodhouse says, "is that the velocity anomalies are due to temperature differences,
but we cannot rule out chemical differences." Many researchers believe that
the major cause of a velocity Increase at
about 450 kilometers is a pressureinduced rearrangement of atoms in
olivine. The rearrangement produces a
denser structure, resembling the mineral spinel. "It is not clear yet," says
Dziewonski, "whether the disappear-
ance of surface characteristics In this
depth range is related to this p h a s e
transformation."
Earthquake waves experience another, even more marked jump In velocity at a depth of 670 kilometers. This
can also be explained as a phase change
in which atoms are squeezed Into a
structure resembling that of the mineral
perovskite. Spinel has each of its silicon
atoms s u r r o u n d e d by four oxygen
atoms; perovskite is even denser, packing six oxygens around each silicon.
A fundamental change
The 670-kilometer discontinuity apparently marks a fundamental change
In the earth. Most if not all earthquake
generation stops at this depth. Seismic
waves from shallower temblors increase
their speed at this boundary, but few if
any earthquakes originate beyond it.
The abrupt cutoff appears w h e n geo-
MOSAIC 33
physicists map earthquakes generated
by the edges of sinking plates. The foci
of these shocks descend at an angle under the overriding plate, then suddenly
stop at 670 kilometers.
The simplest explanation is that the
sinking slab meets resistance so strong
it cannot move deeper. As with most
simple explanations of what goes on in
the deep earth, not everyone agrees
with this one. Geophysicist Thomas H.
Jordan of MIT examined the travel times
of earthquake waves that p a s s e d beneath the Sea of Okhotsk, where a plate
carrying the Kamchatka Peninsula is
plunging under the northeastern part of
Siberia. He insists that these data "require penetration by the Kamchatka slab
to depths of at least 1,000 kilometers,
although penetration d e p t h s much
greater cannot be excluded/' Bradford
Hager, a Caltech geophysicist, takes the
middle road: that the slabs meet resistance but do go through the barrier.
Woodhouse and Domenico Giardini
at Harvard University have studied
earthquakes in the Tonga Island area,
where the Pacific plate sinks under the
Australian plate. They conclude that the
sinking slab breaks up at the 67Q-kilo~
meter discontinuity. "But/' they add,
"the pieces may go through and not
cause deeper quakes because of changes
in material properties with pressure and
temperature."
Existence of this structural or chemical discontinuity is central to the question of whether it prevents movement of
material from the lower to the upper
mantle. The geological community is divided on the issue. There are those researchers who believe that convection
takes place in cells that extend from the
top of the core to the bottom of the
crust. Others maintain that the discontinuity separates convection In the
lower 2,100 or 2,200 kilometers of the
mantle from that In the upper 670 kilometers. They agree that heat moves
across the barrier, but Insist that rockforming materials do not.
Tomographic Images do not resolve
the issue. Surface waves reach the limit
of their resolution at 650 to 700 kilometers, and body-wave data are least
certain at the same depth. "The hot spot
near Hawaii seems to extend through
the barrier," Dziewonski says. "Also, the
low-velocity anomaly associated with
the Red Sea-Gulf of Aden hot spot extends below the discontinuity and appears to merge with other deep, low-
34 MOSAIC
velocity anomalies that continue
obliquely to the core." These data, therefore, do not show evidence of an interruption In mantle convection. However,
frustrating uncertainties in resolution
and interpretation of the data persist,
making it impossible to rule out twolayer convection. (See "Mantle Geochemistry: More Than One Riddle," by
Blanchard Hlatt, Mosaic, Volume 12,
Number 2.)
Another frustration involves the obvious lack of a clear relationship between the lateral extent of postulated
convection cells and motions of the
plates. Cold crust sinks at convergence
zones, but rising hot material is not confined to spreading centers between diverging plates. Convection may not
occur in simple circular or elliptical cells;
rather it may Involve eddies within eddies within even larger eddies.
Once geophysicists accept convection, they face the question of
whether it controls or is controlled by
plate position. In the latter case, insulation of the mantle by the overlying continents causes overheating, which initiates an episode of continental rifting.
When the continents come to rest, the
earth enters a long period of stability,
perhaps without convection. (See
"What Drives the Earth's Plates," by Patrick Young, Mosaic, Volume 10, Number
5.)
If convection controls plate motion, it
may carry the plates along like a conveyor belt that p u s h e s at mid-ocean
ridges and pulls at convergence zones.
Studies of plate motion with respect to
hot spots reveal that continents and
ocean floors drift in a way consistent
with motion away from the ridges and
toward zones of subductlon: Plates with
a large part of their boundary sinking
Into the mantle move most rapidly,
about eight centimeters a year. Those
without substantial subducting borders
poke along at only about two centimeters a year. The slow movers Include all the plates carrying large
continents.
Windows in store
Unknowns about the earth's structure
and dynamics also can be approached
by way of the chemistry of deep rocks
brought to the surface by volcanism and
other tectonic upheavals. Particularly
helpful are studies of Isotopes whose
ratios in the rock are not affected by
high-temperature reactions that alter
materials on their journey u p w a r d .
Such reactions melt or partly melt rocks,
causing separation of light and heavy
fractions. The continents, for example,
were formed from lighter material that
rose to the surface of a partially molten
mantle. Separated fractions, however,
carry isotoplc fingerprints that enable
researchers to trace their origins. In addition radioactive Isotope ratios act as
calendars that date the separations.
Fractionation, for example, concentrated elements such as the rare earth
neodymium and the alkali metal
rubidium In the crust, while the rare
earth samarium and the alkali metal
strontium remained In the mantle. Radioactive samarium 147 decays to neodymium 143, which Is thereby increased
in amount, while the quantity of neodymium 144 remains constant. This activity occurs in both the crust and
mantle, but the ratio of neodymium 143
to neodymium 144 grows more rapidly
in the mantle, because more samarium
is decaying there. Conversely the ratio
of strontium 87 (whose parent is
rubidium 87) to strontium 86 is greatest
in the continents.
Advanced mass spectrometers enable
researchers to measure the ratio of isotopes in a rock relative to their ratio in
the whole, undifferentiated earth. The
whole-earth ratio Is impossible to determine directly, so geochemists use the
average composition of a class of stonv
meteorites, carbonaceous chondrites, as
a reference. Geologists assume that the
planets formed from the same cosmic
materials as these meteorites. The materials undoubtedly were altered during
the formation of the planets, but the
proportions of refractory elements, such
as neodymium and samarium, are assumed to have remained the same.
Rocks from different parts of the crust
show different variations from the
whole-earth standard, Indicating that
they did not come from the same part of
the mantle. This fact supports the Idea
of a layered mantle because, if the whole
mantle Is converting, differences would
be eliminated. Whole-mantle convection, however, may have existed In
the early history of the earth, when the
planet was much hotter. Caltech's Don
Anderson believes that vigorous stirring, produced by more Intense thermal
gradients between the core and the surface, would prevent formation of a permanent crust until the planet cooled
and convection slowed. He cites this as
the reason that no rocks older than 3.8
billion years have been found on a planet estimated to be 4.5 billion years old.
As the earth cooled, low-density fractions of the mantle formed a stable
crust. Geochemists argue that this crust
could have formed from the upper third
of the mantle, leaving the lower twothirds unaffected. Those who believe it
happened this way see the mantle below a 670-kilometer depth as a reservoir
of primitive material that has existed
separately from the upper mantle for
two or three billion years.
Geoalchemy
The separation, however, does not
rule out an exchange of material. Some
earth models call for material to rise
from the lower mantle in the form of
diapirs, or huge blobs, that pass unaltered through the upper mantle and
break through weak spots in the crust.
Massive outpourings of basalt between
15 million and 60 million years ago, in
what now is the Columbia River basin of
Washington and Oregon, for example,
bear isotopic fingerprints similar to
those of reference meteorities.
Stanley R. Hart, who lists his affiliation as "the Center for Geoalchemy,
Massachusetts Institute of Technology/'
maintains that "no primitive mantle ex-
ists/' and no material can make the journey from the lower mantle unaltered.
He believes that mixing in the upper
mantle, as well as reactions with solutions and vapors, produces a blend of
isotopes that sometimes mimic those in
the meteorites.
Many blends exist. Hot spots, which
may extend into the lower mantle, have
isotopic signatures that are different
MOSAIC 35
from the Columbia River basalts and
from each other. Isotopic ratios in volcanic rocks from Hawaii are intermediate between those that erupt along
ocean ridges and those that characterize
a primitive earth. Rocks from Kerguelen
Island in the southern Indian Ocean
show an enrichment of neodymium and
rubidium more typical of the crust than
the mantle. Such confused identities
may result from different depths or origin; they might also derive from combinations of contaminants from the
upper mantle and crust.
Isotopic fingerprints of volcanic rocks
on some but not all island arcs at the
edges of subducting plates are totally
different from those on intraplate islands. The island-arc rocks hold
amounts of strontium typical of continental rocks. Geochemists believe that
water dissolves strontium out of the
continental rocks and carries it to the
sea. Strontium-laced seawater eventually mixes with basalts at the ocean
ridges and becomes incorporated into
oceanic crust. When this crust finally
becomes the sinking edge of a plate, it
mixes with sediments accumulated on
the ocean floor and with mantle material above the sinking slab. This combination melts and erupts in island-arc
volcanoes. Subsequent crustal compression welds t h e s e islands to the continents, creating mountain ranges such
as the Andes and the Cascade Range.
'Island arcs are factories that make new
continental crust/' notes Stanley Hart.
"Fractionation of the mantle above the
sinking plate edge is the sole means of
producing new continental material."
Hart postulates that continents grow
by this means at the expense of oceans.
"An increase in the size of the continents
is balanced by a decrease in ocean-ridge
volume," he says. Geologist William S.
Fyfe of the University of Western Ontario believes just as firmly that the continents are shrinking. He maintains that
the loss of sediments eroded from the
continents and the wearing away of continental edges at subduction zones exceeds the a m o u n t of new crust produced by island-arc volcanism.
Isotope ratios Indicate that much of
the crust separated from the mantle between 3.5 billion and 2.5 billion years
ago. There are, however, no measurements to reveal how the continents
were formed. They could have originated as a thin, globe-covering crust
that broke u p and thickened into con-
36 MOSAIC
tinents. Or they could have grown
around cores that formed over heterogeneities in the mantle, where light fractions rose to the surface.
Once plate tectonics began, however,
the continents were able to grow by collision. Geologists, w h o search out
suture zones that mark small and catastrophic collisions, trace plate tectonics
back 1.3 billion years. N o n e of the
basalts associated with subduction,
mountain building and other tectonic
events are older than one billion years,
and their distribution hints that plate
tectonics Is younger t h a n the continents. Some geologists claim that
these rocks do not occur In the oldest
central core areas of continents; therefore the continents preceded subduction and other events that characterize
plate tectonics. Other geologists claim
that so-called greenstone belts In the
continental cores possess characteristics
similar to volcanic rocks on modern island arcs. William Fyfe, w h o believes
that continents predate plate tectonics,
speculates that hot-spot tectonics dominated the sculpturing of the earth before
plate tectonics.
Temperature and composition
However continents formed and plate
tectonics began, their origins are attributable to the same processes that
drive geological evolution today: density and heat differences. Studies of the
propagation of seismic waves provide a
good look at variations In density, but
temperature gradients are more difficult
to determine. Orson L. Anderson, a
geophysicist at the University of California at Los Angeles, compares our
knowledge of the earth to a three-legged stool. "One leg Is temperature; one,
density; and one, p r e s s u r e / ' he says,
"Density and pressure are known fairly
well, but lack of data about temperatures makes the stool wobbly/
Heat now coming out at the surface
must have originated from the formation of the planet and from the decay of
radioactive elements. The problem is to
determine the relative contribution of
each source. Theories that the earth was
born hot, for the most part, have replaced those proposing that It began
cold. "All the evidence suggests that the
earth accreted after metal and silicate
particles condensed out of a cooling solar nebula/' says Raymond Jeanloz, a
geophysicist at the University of California at Berkeley.
These particles may have come together as a result of gravitational attraction, after which the material differentiated into a dense iron-rich core and lessdense silicate-rich mantle and crust. As
iron particles sank to the center, they
would release enough gravitational energy to raise the earth's temperature
about 1,500 degrees Celsius. Alternatively metallic particles could have agglomerated into a core, which then
attracted silicate materials. If the silicates
accreted rapidly enough, heat would
have accumulated faster than it could radiate into space.
"An Increasing number of scientists
believe that the major source of present
temperature gradients is this primordial
fire/' says Orson Anderson. Again, not
everyone agrees. ''Enough radioactive
elements, such as u r a n i u m , thorium
and potassium, exist in the mantle to
account for more than half of the earth's
total heat loss," Hart says.
Anderson, however, does not believe
that this heat is needed to explain what
goes on. He notes that helium and
xenon gas dating from the earth's birth
have been detected escaping from the
interior through mid-ocean ridges. This
indicates that both primordial gases and
heat have been leaking out of the planet
for 4.5 billion years. As the planet
cooled, core material is believed to have
crystallized, accounting for seismic evidence of a solid core 1,100 kilometers in
diameter and enveloped by a molten
outer core. As the outer core loses its
heat through the mantle, the Inner core
slowly increases In size. The two parts
can be seen clearly through the seismic
window, because compressional waves,
which vibrate in the direction of travel,
become severely attenuated by molten
or liquid material.
The velocity of seismic waves is consistent with the theory that the inner
core consists mainly of Iron. Calculations of the earth's moment of inertia
and of the cosmic abundance of elements, as well as the need for a metallic
core to account for the planet's magnetic
field, also suggest an iron core.
Determining core composition is vital
to putting knowledge about the earth
on a stable footing, because the temperature of a molten outer core depends
on the material of which it Is made. This
temperature then becomes a reference
for calculating the planet's heat balance.
J. Michael Brown of Texas A. & M.
.University and Robert G. McQueen of
the Los Alamos National Laboratoryhave attempted to put an upper limit on
this temperature by duplicating pressures and materials believed to be
characteristic of the core. They compressed pure iron to a pressure of 2.5
megabars, or millions of atmospheres,
and measured its melting temperature
as 5,000 degrees Kelvin. By extrapolation they calculated that temperatures
where the inner and outer cores meet
reach about 5,800 degrees. "This is the
first time temperature estimates near
the earth's center have been obtained by
experimental data," says Jeanloz.
Brown and McQueen conclude that
the inner core is principally pure iron
and that the outer core contains an inhomogeneous mixture of iron alloyed
with lighter elements. The lighter elements must be present, because calculations of the earth's density are inconsistent with a core made entirely of iron.
Moderate amounts, some 8 to 10 percent by weight, of such elements such
as oxygen, silicon, or sulfur must be
present to produce a match with seismic
and astronomical observations.
Combining iron with a lighter element, however, lowers its melting
point. Sulfur, for example, has a much
larger effect on the melting point than
oxygen does at lower pressures; if this
difference persists at high pressures, an
outer core with iron sulfide would be
much cooler than one with iron oxide.
The uncertainty about composition introduces an uncertainty of about 1,000
degrees Celsius in the temperature at
the core-mantle boundary.
Diamond window
The next step for laboratory scientists
is to determine the melting points of different alloys at pressures between 1.5
million and 3.5 megabars to see which
melting point fits best as the third leg of
the stool. Researchers can do this by
squeezing a microscopic sample of material in a so-called diamond anvil cell.
In such a cell, the sample sits between
two flawless, flat-faced diamonds driven
together by a screw mechanism attached
to a sliding piston. By improving on this
simple device, geophysicists at the Carnegie Institution of Washington have
produced a pressure of up to 2.8 megabars in a chamber 200 microns in diameter and 50 microns in height (a feat that
is included in the Guinness Book of World
Records). Laser beams flashed through
diamond windows in the cell can heat
Bell. Guinness-record pressures.
samples to 5,300 degrees Kelvin. "With
such experiments, which touch the limit
of technology, we are starting to duplicate conditions at the core-mantle
boundary," says Carnegie's Peter M.
Bell. "I have no doubt that in the near
future we will be able to reach pressures
and temperatures as high as those at
earth's center."
Such experiments already have
changed established views of the planet's interior. Before diamond-cell work
began in the early 1960s, geophysicists
believed they could not account for density differences in the deep earth simply
by putting crustal granites and basalts
under high pressure. "Diamond-cell experiments at pressures of 200 kilobars
and 1,500 degrees Celsius proved this
wrong," Bell says. Under these conditions, believed to prevail at the 670-kilometer discontinuity, minerals found at
the surface take on the crystal structure
of perovskite,. the density of which can
account for the behavior of seismic
waves in the lower mantle. "Silicateperovskite is the densest known mixture of crustal silicates and oxides," the
Carnegie scientist points out. "Such a
structure may exist all the way to the
core-mantle boundary."
No one knows what happens at this
boundary. Seismic data indicate a
mushy, semifluid zone of reduced velocity similar to what would be expected
from a slurry of crystals and liquid.
"Solid-liquid phase transitions like this
normally are accompanied by changes
in chemical composition," Bell says. "It
is difficult, however, to determine
whether physical or chemical change
dominates."
Bell and his colleague Ho kwang
(David) Mao propose that chemical reactions predominate, kicking iron out of
the mantle and into the core. This concept, they speculate, could account for
the presence of Iron oxide in the outer
core. If phase change dominates, silicates could behave like metals because
of the increased pressure. This concept
raises the possibility that the core may
contain other compounds besides Iron
alloys. Raymond Jeanloz is using shock
waves and diamond anvil cells to study
the possibility that calcium oxide exists
in a metallic phase in the core. Researchers have not found the amount of
calcium that they were led to expect by
its abundance in stony meteorities.
"This missing calcium may have been
squeezed into a super-dense phase and
stored in 'the core," Bell suggests.
Diamond anvil cells, like other windows to the earth's interior, Impose new
limits on what Is possible and so constrain speculation. "Only 25 years ago,
there were so many theories that placing
your belief was like an act of faith," says
Robin Brett of the U.S. Geological Survey. "Many different faiths coexisted
because none could be proved or disproved." The plate-tectonics model destroyed many of these faiths and helped
to unify those that survived.
Brett says plate tectonics theory
"sparked a revolution in geology that
has been as important as the theory of
the atom was to physics and the theory
of evolution to biology." The theory,
however, successfully accounts for the
development of only the youngest twothirds of the earth's surface, the ocean
basins. The technologies that are opening new observational windows are extending the revolution to the continents
and to more distant time periods.
Of course answers to fundamental
questions still He In the future, as does
the application of the revolutionary
knowledge to finding mineral and energy resources, predicting earthquakes
and volcanoes, and forecasting sea-level
changes. Nonetheless, these new windows enable scientists to look at the
earth with more Insight and to predict
its future with more certainty. •
The National Science Foundation contributes
to the support of research discussed in this
article through programs in its Division of
Earth Sciences.
MOSAIC
37